Full text of "USPTO Patents Application 10687850"

See other formats

per

WORLD INTELLECTUAL PROPERTY ORGANIZATION
International Bureau

INTERNATIONAL APPLICATION PUBLISHED UNDER THE PATENT COOPERATION TREATY (P CI)

(1 1) International Publication Number: WO 92/01928

(43) International Publication Date: 6 February 1992 (06.02.92)

(51) International Patent aassification 5
G01N 27/30, 31/12, 27/02
G01N 27/416, 27/28, 27/00
COIN 31/00, G01F1/66

(21) International Application Number:

PCT/US91/0507I

(22) InterarSonal Filing Date: 18 July 1991 (18.07.91)

20 July 1990 (20.07.90) US

(30) Priority data
555,976

(71) Applicant: I-STAT CORPORATION [US/US]; 303 Col-
lege Road East, Princeton, NJ 08540 (US).

(72) Inventors: COZZETTE, Stephen, N. ; 45 Dennis Court,
Hjghtown, NJ 08520 (US). DAVIS, Graham ; 15-04 Fox
Run Drive, Plainsboro, NJ 08536 (US). HOLLERIT-
mS* 7^T^ n i 14 Sunset Unt > 0ak Wd » NJ 07438
£3. R ' ; 10,1 Yardley-MorrisviHe
Road, Yardley, PA 19067 (US). PIZNIK, Sylvia ; 12

, nn o.o owt » Jackson > N J 08527 (US). SMIT, Nicola-
llin fh^^Yr^\ tn ^ Woodkwn. Ontario K0A
3M0 (CA). TIRINATO, Jody, Ann ; 27-06 Hunters Glen

?i r S^ P,a, n nsb S r S!r 08536 < US >- mEC *> Henry, J. ;
31 Parker Road, Piamsboro, NJ 08536 (US). ZELIN, Mil
chad, P. ; 9104 Tamarrbn Drive, Plainsboro, NJ 08536
(US).

(74) Agent: MISROCK, S., Lesbe; Pennie & Edmonds, 1155
Avenue of the Americas, New York, NY 10036 (US).

ffi C ^ CH (European patent), DE (EuropeanV
ent), DK (European patent), ES (European patent), FR
^uropean patent) GB (European patent), GR (Euro-

pean patent), NL (European patent), SE (European pa-

Published

With international search report
Before the expiration of the time limit for amending the
claims and to be republished in the event of the receipt of\
amendments, 1

' (54)TMe: METHOD FOR ANALYTICALLY UTILIZING MICROFABRICATED SENSORS DURING WET-UP

(57) At act

first time window in the presence of the i«S KZ„1^ ^L?k $ Pfrf»nnu,g a first signal measurement in a
ment; (I) contacting the ^SS^SS^SS^^t ftid 1 T ? e first signal measure -

ond time window in the presence of the^nleS a^M ^ performing a second s.gnal measurement in a sec

FOR THE PURPOSES OF INFORMATION ONLY

Codes used Co identify States party to the PCT on the front pages of pamphlets publishing international
applications under the PCT.

Austria

Spain

Madagascar

Australia

Finland

Mali

Barbados

France

Mongolia

Belgium

Gabon

Mauritania

Burkina Faso

United Kingdom

Malawi

Bulgaria

Guinea

Netherlands

Bcnio

Greece

Norway

Brazil

Hungary

Poland

Canada

Italy

Romania

Central African Republic

Japan

Sudan

Congo

Democratic People's Republic

Sweden

Switzerland

or Korea

Senegal

CStc dlvoire

Republic of Korea

su*

Soviet Union

Cameroon

1 ittfcfrtcnstcin

Chad

Czechoslovakia

Sri Lanka

Togo

Germany

United States of America

Denmark

Monaco

+ It is not yet known for which States of the former Soviet Union any designation of the
Soviet Union has effect.

WO 92/01928

-1-

PCIYUS91/05071

METHOD FOR ANALYTICALLY UTILIZING
MI CROFABRI CATED SENSORS DURING WET-UP

CROSS-REFERENCE TO RELATRn APPLTPATTOWg

5 The present U. S. Application is related to

prior co-pending U. S. Application Serial Nos. 07/432,714,
filed November 7, 1989, and 07/245,102, filed September 15,
1988, the disclosures of which are incorporated by
reference herein in their entirety,

10 1 . FIELD OF THE INVENTION

The present invention relates to a method of
quantifying a preselected analyte species present in
fluids, which takes advantage of the well-behaved
equilibration wet-up characteristics of dry-stored wholly

15 microfabricated electrochemical sensors. These and other
performance characteristics obtain from a manufacturing
process, described in the co-pending U.S. Application
Serial No.. 07/432,714, that attains a high degree of
uniformity with regard to the physical dimensions and

20 resulting ; properties of such electrochemical sensors. The
invention allows for the acquisition of analytical data
while the signal of the sensor of interest is still
undergoing the process of an equilibration wet-up. In
particular, the present method includes deriving useful

25 information from microfabricated electrochemical sensors,
which sensors had been stored dry, much more quickly than
previously thought practical by acquiring and manipulating
selected signal measurements well before the sensors have
attained a post-equilibrated wet-up state (i.e., steady-

30 state) response. Most notably, the invention provides a
method for relating the signal measurements recorded in
different fluids to determine the ratio of the

WO 92/01928

- 2 -

PCT/US91/05071

concentration of a preselected analyte species in each
fluid. The present method utilizes a computational means
which is able to distinguish the signal response of a given
electrochemical sensor to changes in analyte concentration,
5 which response is fast relative to the slower monotonic
"wet-up" behavior of the sensor and its~ associated
reference electrode.

2 . BACKGROUND OF THE INVENTION

The recent emphasis in the development of

10 clinical chemistry technology has been directed toward the
development of systems for "real time" analysis of
biological fluids or those analyses which can be performed
in the close proximity of the patient e.g., at the bedside
or in the physician's office. Such biological fluids

15 include urine, plasma, serum, and preferably, whole-blood.
Clear benefits are . achieved if the chemical information
required by the physician is obtained during patient
consultation and not several hours or days afterward.
Although progress has been made toward achieving such a

20 goal, many problems still remain including the limitations
of established manufacturing methods to mass-produce
electrochemical devices with sufficiently uniform
performance characteristics and extended shelf-lives. Of
particular interest is the lack of adequate computational

25 techniques which minimize the time required to obtain
useful information from such electrochemical devices.

To date, fluid analysis has been carried out
using many types of electrochemical sensors in which
potentiometric, amperometric or conduct imetric
30 measurements are performed in a steady-state or kinetic
(e.g., initial rate) mode. Electrochemical sensors
employed for these measurements usually consist of e two-
component assembly in which a sensitized membrane is
interposed between the fluid and an underlying electrode.

WO 92/01928

- 3 -

PCT/US91/05071

Some membrane compositions have distinctive species
recognition capabilities which enable the electrochemical
sensor to detect the analyte of interest specifically and
measure its concentration in a complex biological fluid.
5 To date, however, size, complexity and expense limitations,
combined with a high incidence of error resulting both from
instrument quality control and accidental errors by the
operator, have impeded wide spread use of this technology
in locations, such as the emergency room and the doctor's

10 office, which are remote from the central clinical
chemistry laboratory. Moreover, most analytical test
methods currently in use are overly cumbersome or complex.
More significantly, the response of the electrochemical
device, itself, may be so slow as to make such "real time"

15 analysis very difficult. It should be noted that the
concentration of certain components, such as glucose and
potassium ion, in a biological fluid (e.g., whole-blood)
may change significantly over a prolonged period. The
change arises likely from hemolysis "of related metabolic

20 processes.

As mentioned previously, a principal obstacle
against the successful implementation of "real time"
clinical fluid analysis is the lack of reliable sensor
manufacturing methods. Equally lacking, however, are data
25 acquisition and manipulation methods which allow the quick
retrieval of information from existing chemical sensing
devices some of which are stored substantially dry in order
to maximize shelf-life. The prevailing standard practice
dictates that these "dry-stored" devices be allowed to
reach a fully equilibrated "wet-up" state before meaningful
sensor data can be recorded.

WO 92/01928 PCT/US91/05071

- 4 -

2.1. PREVIOUS DEVICES AND METHODS
FOR FLUID ANALYSIS

Some progress has been made toward the
production of improved testing apparatus, including
5 miniaturized chemical sensing devices. U. S. Patent No.
4,734,184 issued to Burleigh et al. discloses an electrode
assembly for monitoring the concentration of a number of
gases and ions present in the blood. Although the assembly
is stored dry to promote an extended shelf-life, the

10 electrodes, are thoroughly hydrated (wet-up) prior to use.
During operation they are in prolonged and equilibrated
contact with the many solutions employed/ including a
calibrant solution, a reference solution and intermittent
blood samples. Thus, during the continuous monitoring, for

15 up to 36 hours, of a subjects blood gases, electrolytes
and hematocrit levels, all measurements are performed with
the sensors providing signal responses in the steady-state.
No disclosure is included for deriving meaningful
analytical information from solid-state electrodes before

20 the electrodes attain an equilibrated "wet-up" state.

U. S. Patent No, 4, 654, 127- issued to Baker and
Funk discloses a sensing device equipped with species
selective sensors and a rotatable multichamber reservoir in
which calibrant and sample solutions are contained but in

25 separate chambers. A plurality of chemical species may be
detected by this device. However, as the sensors employed
are not microf abricated, only a limited amount of control
over the dimensions of the sensors ■ various components had
been possible, resulting in their having nonuniform

30 response behavior and necessitating the batch-wise
determination of each sensor's response. The value of the
response, e.g., intercept and slope, is then recorded on a
bar code which must be read by a table top analyzer before
the concentration of the desired chemical species may be

WO 92/01928

- 5 -

PCT/US91/05071

calculated. Moreover, the disclosure of this patent is
silent on methods for making useful measurements during
wet-up of the sensing device. All indications, including
those from available product literature, support the
supposition that the- calibrant and sample solution
measurements are carried out after the sensor response has
attained a steady-state value. Furthermore, these
commercially available sensors are stored in a high
humidity package (i.e., substantially wet). This packaging
method has the effect of limiting the shelf-life of these
sensing devices significantly compared to competitive dry-
reagent systems, such as those described below. However,
this compromise ensures that the device is substantially
"wet -up" at storage and, thus, enables the sensor to
15 provide results fairly rapidly. Unfortunately, this
compromise cuts back the device's useful shelf -life quite
severely, particularly for enzyme-based sensing devices.
Shelf-life, of course, can be extended to some extent by
refrigerating the package. However, refrigeration adds to
the expense of storage and also means that the device must
be allowed to return to room temperature before use.

U. S. Patent Nos. 4,708,776 and 4,608,149
disclose, on the other hand, improved "dry-operative" ion-
selective electrodes. The inventors describe a " "dry-

25 operative" electrode as "an ion-selective electrode which
provides reproducible potent iometric determination [of] ion
activity which is related to the ion concentration of an
aqueous test solution with no requirement for wet storage
or preconditioning prior to use" (col. 2, lines 10-15 of the

30 second patent listed above). These patents also disclose
methods of using such electrodes. In particular, the
pbtentiometric determination of the concentration of sodium
and potassium ions in an aqueous liquid is described.
However, the method relies on a differential measurement

35 which involves contacting the first of two "uniform"

WO 92/01928

- 6 -

PCT/US91/05071

electrodes i.e., a pair of identically formed potassium
ion-selective electrodes, with the sample liquid and
contacting the second electrode with a reference liquid
(calibrant) containing a known amount of the ion of
5 interest and then determining the the resulting difference
between the two potential readings. This method requires
that the sample and calibrant solutions be brought into
contact with the respective electrodes strictly
simultaneously to obtain reliable measurements of the
10 analyte concentrations. Consequently, it is necessary to
provide an automated means for the simultaneous application
of the calibrant and sample fluids to prevent errors in the
measurement .

Pace, in European Patent Application No. 0 012

15 035, describes self-calibrating miniaturized multiple
sensors fabricated on a single chip. The usefulness of
this disclosure is quite limited as the exact nature of the
materials used for each of the multitude of layers
described in the complex sensor structures is not revealed.

20 Pairs of identical electrodes are used, a first member of
the pair having at least two distinct electrolyte "layers"
of known composition and the other, member of the pair
either having no electrolyte present in its corresponding
layers or having electrolyte present therein at a

25 concentration which is significantly different from the
first member. A discussion of the self-calibrating nature
of these pairs of matched electrodes is present and makes
clear that a differential method of signal measurement is
employed to "nullify any drift and offsets in the

30 measurement" (page 23, lines 30-31) m Moreover, this
reference asserts further that these multiple layers which
provide self-calibration "not only assure built-in
reliability and accuracy, but relax manufacturing
tolerances" (page 26, lines 1-4) . Thus, no successful

35 means has been disclosed to manufacture simpler structures

WO 92/01928

- 7 -

PCT/US91/05071

with high dimensional tolerances nor has there been any
suggestion that useful information may be derived from
signal response measurements prior to attaining complete
wet-up, much less that a method may be formulated in which
5 such measurements are exploited.

Accordingly, there remains at the present a need
for a method which integrates a sensing device, preferably
a microf abricated electrochemical sensor, having the
requisite predictable, reproducible chemical response and

10 "wet-up" characteristics, and an effective computational
technique, which method allows the physician to obtain
conveniently precise, accurate determinations of the
concentration of analytes of clinical interest. Such
determinations are desirably made in five minutes or less,

15 most preferably within about a minute.

2.2. PREVIOUS USES OF POTENTIAL PULSES

Previous workers have utilized potential pulse
techniques to increase the sensitivity of the
elec chemical measurement or to reduce the flow

20 dependence, of the electrode signal. However, these
previous "applications have always involved fully wet-up
devices utilized for the continuous monitoring of analyte
concentrations. In such applications, previous workers
sought to improve the signal output by taking their

25 readings immediately after the application of a potential
pulse across the sensor 'ectrodes. The cathode is^ open-
circuited between* the pulses of applied potential . These
and related techniques are described more fully in Short,
D. L. and Shell, G. S. G. J. Phvs. E.?SrH ... Tnatnim. 19B5,

30 13_ r 79-87 and Lilley, M. D. et al. J. Elertr oanal. Chpm.
1969, 21, 425-429.

On the other hand, methods exist for activating
a catalytic surface including polishing away the surface

WO 92/01928

- 8 -

PCI7US91/05071

layer with a fine particle size, inert abrasive material
such as alumina or placing the electrode in a corrosive
acidic solution such as 1 molar sulfuric acid and cycling
the applied potential for several minutes. Clearly, these
5 existing methods are destructive and are inappropriate for
the activation of an electrode surface overlaid with
microf abricated biolayers .

3 . SUMMARY OF THE INVENTION

In accordance with the present invention a
10 method is disclosed for determining the ratio (s) of the
concentrations of preselected analyte species in more than
one fluid which comprises, in part, providing
microf abricated sensors having the requisite
characteristics, which will be described more fully below,
15 and performing signal measurements , before the equilibrated
wet-up process is complete, while the sensors and reference
electrode are in ; contact with a first fluid and,
subsequently and separately, with a second fluid.

In the present method, which fluid is brought
20 into contaQt with the sensor first is unimportant, so long
as a separate signal measurement is made while each fluid
is in contact with the sensor and reference electrode.
Despite the fact that, such measurements are taken before
the sensors are fully w wet-up, n a process which may take
25 several minutes, useful analytical, information about a
variety of biological analytes can still, surprisingly, be
obtained. The computational techniques for extracting this
information from dry-stored microf abricated electrochemical
sensors are disclosed. Thus, the present invention allows
30 the simplicity and dry-storage capabilities of
microfabricated sensors to be exploited while providing
measurements of preselected analyte species as close to
"real time" as possible.

WO 92/01928

- 9 -

PCT/US91/05071

It is thus an object of the present invention to
provide a method of determining the concentration ratio of
a preselected analyte ' species present in at least two
fluids comprising (a) providing at least one
5 microfabricated chemical sensor, which exhibits a response
to changes in the concentration of a preselected analyte
species, and a reference electrode capable of sustaining a
well-behaved reference potential for a period of time
sufficient to permit the completion of at least two signal

10 measurements, against which reference potential of said
sensor is measured, which sensor and reference electrode
have been stored substantially dry, and which response is
sufficiently rapid or "fast" relative to the "slow"
monotonic wet-up behavior exhibited by said sensor and

15 reference electrode when contacted with fluid; (b)
establishing electrical contact between said sensor,
reference electrode and external computational means; (c)
contacting said sensor and reference electrode with a first
fluid; (d) performing the first of said signal measurements

20 in a preselected first time window in the presence of said
first fluid; (e) displacing said first fluid; (f)
contacting* said sensor and reference electrode with a
second fluid; (g) performing the second of said signal
measurements in a preselected second time window in the

25 presence of said second fluid; and (h) relating said first
and second signal measurements to the known concentration
of said analyte species in one of said first or second
fluids, to determine the unknown concentration of said
analyte species in the other of said fluids before said

30 sensor attains full equilibrated wet-up.

It is also an object of the prese— invention to
provide a method of determining the concentration of a
plurality of preselected analyte species present in a
sample fluid comprising, as a first step, providing an
35 array of microfabricated potentiometric and amperometric

WO 92/01928

- 10 -

PCT/US91/05071

sensors, each sensitive to changes in the concentration of
a particular preselected analyte species, and which array
also comprises one or more reference electrode capable of
sustaining a well-behaved reference potential for a
5 sufficient period of time. Preferably, one reference
electrode is dedicated to said potentiometric sensors and
an other reference electrode is dedicated to said
amperometric sensors. However, a single reference
electrode may also be used for both types of sensors, if

10 said amperometric sensors are also supplied with a common
counter electrode. The counter electrode is designed to
prevent polarization of the reference electrode, the effect
of which polarization is more deleterious to the
performance of potentiometric sensors. As noted, above,

15 each sensor has been stored substantially dry and exhibits
a response to said changes in the concentration of said
particular preselected analyte species which is
sufficiently rapid relative to the monotonic wet-up
behavior of said sensors; (b) establishing electrical

20 contact between said array of sensors and external
computational means; (c) contacting said array of sensors
with a first (e.g., calibrant) fluid; (d) performing a
first set of signal measurements in 'a preselected first
time window in the presence of said first fluid; (e)

25 displacing said first fluid; (f ) contacting the array with
a second (e.g., sample) fluid suspected of containing said
plurality of analyte species, such that said array of
sensors is in contact with said second fluid; (g)
performing a second set of signal measurements in a

30 preselected second time window in the presence of said
second fluid; and (h) relating said first and second sets
of signal response measurements to determine the
concentration of a plurality of said preselected analyte
species in said second fluid, based on the Lnown

35 concentrations of each of said preselected analyte species
in said first fluid.

WO 92/01928

- 11 -

PCT/US91/05071

It is a further object of the present invention
to provide a method for activating the electrode surface
of an amperometric sensor by subjecting the sensor to a
series of potential changes while it is in contact with a
5 fluid, preferably a calibrant fluid.

Yet another object of the present invention
involves providing a conductivity sensor by which the
conductivity of the fluid in contact with the conductivity
sensor may be determined and related, if desired, to the
10 hematocrit level in the sample, or more simply to provide
an indication of whether the fluid is calibrant, plasma,
sr -urn or whole-blood, or even to provide a check on the
pi sence of a sample of calibrant fluid.

Still another object of the present invention
15 includes a determination of the concentration of the
analyte species of interest in about one mir e using dry-
stored sensors .

Yet another object of the present invention
relates to minimizing the incidence of a test "failures" by
20 incorporating data collection methods which allow the
testing apparatus to scrutinize the integrity of the
acquired signals and to manipulate the data set to exclude
extraneous or aberrant data points which may otherwise lead
to a rejection of a particular analysis.

25 Other objects of the present invention should be

readily apparent to those skilled in the art from the
preceding discussion, as well as the following additional
detailed disclosure.

4 . BRIEF DESCRIPTION OF THE FIGURES

W FIG. 1 illustrates the raw waveform exhibited by

a microfabricated potentioroetric potassium ion sensor on

WO 92/01928

- 12 -

PCT/US91/05071

grounding (a) , exposure to a calibrant fluid (b) , initial
sensor wet-up (c) , followed by transition to a sample fluid
(e) . Suitable preselected time windows for data
acquisition in each fluid are indicated by variable time
5 segments (d) , for the first fluid, and (f ) , for the second
fluid.

FIG. 2a-2e show the response to a fluid change
of a potassium ion sensor, sodium ion sensor, chloride ion
sensor, urea sensor and microfabricated on-board reference
10 electrode, respectively, with respect to an external
standard Corning reference electrode.

FIG. 3a-3d show the response to a fluid change
of a potassium ion r sodium ion, chloride ion and urea
sensors, respectively, with respect to a microfabricated
15 on-board reference electrode.

FIG. 4 §hows the signal output (current in
namps) of an amperometric glucose sensor, suitable for use
in the present method, in response to the applied electrode
potential <mV) using a 20 mM glucose in HEPES buffer sample
20 (O) or a HEPES buffer only (X) .

FIG. 5a shows the increasing current output of a
microfabricated glucose sensor as a series of activating
pulse groups or potential changes are applied.

FIG. 5b illustrates the response to a fluid
25 change of the activated glucose sensor of FIG. 4a.

FIG. 5c shows the absence of any substantial
change in the current output of the activated glucose
sensor of FIG. 4a upon application of additional pulse
groups .

30 FIG. 6 illustrates an electrochemical creatine

kinase enzyme assay based on initial rate measurements.

W^92/01928

- 13 -

PCT/US91/05071

This figure also shows the sensor response to a fixed
concentration of adenosine triphosphate (ATP), the product
of the enzymatic process.

5. DETAILED DESCRIPTION OF THE INVENTION

The method of the present invention seeks to
integrate the quantifiable, predictable behavior of
microfabricated sensors during the equilibrated wet-up
process with computational techniques, including data
handling or collection methods, which may be implemented by
the testing apparatus or external computational means, to
arrive at a close estimate of the concentration of an
analyte of interest in a given sample fluid.

The "equilibrated wet-up" process is the means
by which dry-stored sensors, by exposure to a fluid
comprised of an aqueous medium, wet gas and the like, reach
an operational state and, eventually, a steady-state.
Here, the term "equilibrated wet-up" is used to enc.npass,
not only ingress of water through the various membrane
layers to the electrode surface, but also all of the
physicochemical changes that occur prior to a sensor
attaining that steady-state response. These changes and
their consequences include: the hydration of each membrane
layer and its effect upon migration therethrough of
analytes, cof actors, ionopbores, enzyme, affinity-labels,
25 and the like; the hydration of the enzyrae-cbntaining layer
and its effect on the activity of enzymes or the
selectivity of ionophores and affinity labels (i.e., the
effect on their binding coefficients); the hydration of the
electrode surface and the responsiveness of the sensor
30 which may -be a function, for example, of the relative
surface populations of metal oxide and metal hydroxide
sites, or, alternatively, the degree of ligand-ligand
substitution in which a silver-silver halide surface may be
transformed into a halo-aquo metal complex. All of such

WO 92/01928

- 14 -

PCT/US91/05071

changes that occur, before a fully wet-up, equilibrated
sensor is obtained, can give rise either singularly or in
concert, to a monotonically changing signal output despite
the fact that the physical and chemical properties of the
5 fluid, which may be comprised of an aqueous liquid or a wet
gas, in contact with the sensor are not changing (i.e., the
concentration of the preselected analyte in the fluid
remains constant, along with the temperature, pH, osmotic
pressure, ionic strength, etc., of the fluid, especially
10 the calibrant fluid) .

The processes described above may be considered
passive in the sense that upon contact with fluid they
occur spontaneously- However, the electrode surface of an
amperometric sensor presents a special case in that its

15 redox properties are substantially affected by the voltages
that have been applied to it previously. In this sense,
electrochemical activation of the electrode surface by
applying a sequence of difference voltages can contribute
significantly to the reduction of the time it takes before

20 the sensor operates reliably in a steady-state manner, and
therefore, contributes significantly to the apparent "wet-
up" (or RC wet-up) of the sensor.

The monotonic wet-up signal is usually described
by a resistance capacitance (RC) time-constant because it
25 can be simulated electronically by connecting, a resistor
and capacitor either in series or in parallel. Thus, for a
potential step applied to such a circuit, an exponentially
decaying current, is obtained with a time-constant, T= RC
where the current, i - E/Re - (t/t) . In this example the

30 current necessary for changing the capacitor drops to 37 %
of its original value at t ■ T and to 5 % of its original
value at t * 3t. This latter value is often referred to as
the 95 % response-time. For the sensors described here
useful analytical information is obtained well before the

35 sensors have transformed from the dry-state to a hydrated

WO 92/01928

- 15 -

PCT/US91/05O71

or wet-up state in which the sensors provide a signal which
is 95 % of the expected equilibrated steady-state value.

A good introductory discussion of this RC
concept can be found in Lindner, E. et al. "Dynamic
5 Characteristics of Ion-Selective Electrodes," CRC Press,
1988 and the references cited therein, the complete
disclosures of which are incorporated herein by reference.
It should be noted that while the exponential model for
time evaluation of the chemical sensor signal is usually
10 referred to as ah "RC time-constant," no special detailed
elucidation or assumptions about chemical or physical
capacitive or resistive elements of the sensor's operation
are required to verify the applicability of the exponential
model.

15 In the present invention, it has been

surprisingly discovered that the RC time-constant for wet-
up associated with the present wholly microf abricated
sensors and reference electrodes can be manipulated and
modeled closely. The reproducibility and predictability of

20 this, wet-up RC time-constant is, in turn, a product of the
microfabrication techniques described in the following
section, and more fully in related co-pending U.S.
Application Serial Nos. 07/432,714 and 07/245,102. Such
microfabrication techniques provide much finer control over

25 the dimensions of overlaid layers than can be achieved by
previous techniques, including lamination. Indeed, a
multiplicity of electrochemical sensors can now be
fabricated on a single silicon wafer. These
electrochemical sensors, including the first practical

30 microfabricated reference electrode, possess sufficiently
well-behaved properties to allow electrochemical
measurements to be made with the precision and accuracy
required in clinical chemistry.

WO 92/01928

- 16 -

PCT/US91/05D71

As will be described in more detail, below, the
present inventors have discovered that the time evolution
or rate of change in the monotonic wet-up signal can be
predicted to a high degree of accuracy at any time after
5 initial contact of the sensor and reference electrode with
a fluid. This ability to model, analyze and manipulate the
RC time-constant is an important aspect of the present
method which allows the concentration of an analyte species
of interest to be determined quickly and reliably before
10 the sensors have attained a fully wet-up, equilibrated
state.

An equally important element of the present
invention and one which is a direct consequence of having a
reproducible, predictable wet-up RC constant is that the RC
15 constant associated with the sensor's response to changes
in the concentration of a preselected analyte is also
highly predictable and precise.

The RC time-constant, &F, is in units of seconds
(sec) as evident from the following relationships:

20 R (resistance) » £2, ohms » V/i

C (capacitance in F, Faraday) - q/V - (i x sec)/V

since, RC » flF, then

RC = (V/i) x (i x sec)/V .» sec

When a chemical sensor undergoes a change in its
25 signal in response to a change in the analyte
concentration, the exponential time constant governing the
time dependence of the signal of a chemical sensor varies
as taL 2 /D. fsee . Lindner, E. et al. y above, "Dynamic

Characteristics of Ion-Selective Electrodes 11 , DRC Press,
30 19B8 and earlier work by Buck, R.P. cited therein,
especially Chapter 1 of "Ion-Selective Electrodes in

WO 92/01928

- 17 -

PCT/US91/05071

Analytical Chemistry", Freiser, H., (Ed), Plenum Press,
1978).

The time taken by a sensor to attain the steady-
state response to a change in concentration of an analyte
5 can be described by an RC time-constant. Thus, the
RC r esponse time can be approximated by the relationship, RC =
L 2 /D where, L is the thickness of the membrane layer in cm
and D is the diffusion coefficient in cm 2 sec* 1 of the
analyte through the membrane. The diffusion coefficient

10 for glucose through a membrane layer is typically about 100
times less than its diffusion coefficient in solution
(Dsoln * 10" 6 cm 2 sec" 1 , D mem b * iO" 8 cm 2 sec* 1 ) , Thus, if
the membrane layer is about 1 \im (10~ 4 cm) in thickness,
then RC r esponse time is approximated to be (10- 4 ) 2 /10" 8 or

15 ca. 1 sec. The RC W et-upr whose magnitude is on the order of
tens of seconds, is thus "slow" relative to the response
time. For example, the wet-up time-constants for the
sensors shown in the Figures (e.g., FIGS. 1-3) exhibit X

values of ca. 20 sec. As described below, when the wet-up
20 process proceeds to an incomplete level where t<t<3t

i.e., the rate of change in the signal is sufficiently
small, it is possible to model this rate as a linear drift
rate (S££, Table 1) .

Clearly, the time required for a diffusion front
25 of an analyte species to penetrate a membrane layer and
establish a steady-state response increases with increasing
membrane thickness. Thus, having a manufacturing method
which attains a high degree of dimensional control over
sensors 1 overlaid structures is crucial to obtaining
30 devices with predictable, reproducible wet-up and response
time behavior.

In addition to describing a method involving a
single sensor, the present invention is also concerned with
the effective operation of an array of sensors each

WO 92/01928

- 18 -

PCT/US91/05071

sensitive to a particular preselected analyte species.
Each sensor in the array may be operating in one of a
number of possible modes including, but not limited to,
those designed to perform standard potentiometric,
amperometric and conductimetric measurements or kinetic
measurements based thereon. The concurrent operation of an
array of different electrochemical sensors composed of
some, or all, of these types, presents its own unique set
of problems .

Moreover, certain types of microelectrode
assemblies are susceptible to inactivation, or a loss of
surface catalytic activity, and require reactivation to
secure the highest level of sensitivity.

The following is a detail description of each
15 element of the present analytical method for deriving the
concentration of at least one, preferably a number, of
preselected analyte 'species .

5.1. WHOLLY MICROFABRICATED SENSORS

Wholly microfabricated sensors, the availability
20 of which comprises a preferred element of the present
method, are described in detail in the applicants' prior
co-pending U. S. Application Serial No. 07/432,714.
Additional aspects related to the manufacture of integrated
ambient sensing devices, including a microfabricated
25 reference electrode, are described in US Patent No.
4,739,380 and prior co-pending U. S. Application Serial No.
07/156,262, the complete disclosures of which are
incorporated herein by reference. These microfabricated
sensors are manufactured in such a way as to avoid the
30 errors and non-uniformity introduced by manual deposition
of membranes and the like at various stages of the
manufacturing process. Thus, a combination of. thin film
techniques, including wafer-level photolithography and

WO 92/01928

- 19 -

PCT/US91/05071

automated microdispensing, are used to produce hundreds of
identical sensors, or an array of different sensors, on a
single silicon wafer. Reference electrodes made by the
same process are also established in a highly controlled
5 fashion.

Such a reference electrode is necessary for
electrochemical measurement of chemical or biochemical
species in a sample solution ( See r Ives, D.V.G. and Janz,
G.J. "Reference Electrodes, Theory and Practice, " Academic

10 Press, 1961, the complete disclosure of which is
incorporated herein by reference) . In the case of
potent iometric measurement, the signal measured . the
potential of a chemically responsive electrode (s^sor)
with respect to the potential of the reference electrode.

15 Ideally, the potential of the reference electrode is
strictly independent of the chemical composition of the
solution that it contacts. A reference electrode is also
necessary for an amperometric measurement because it
controls the potential of the amperometric sensor. Because

20 such a high degree of control is present with regard to the
composition of these layers, their physical dimensions, as
well as their location on the .sensor array, the
characteristics and specification of each sensor on the
wafer, or any similarly produced wafer, are well-behaved

25 and predictable.

In particular, the microf abricated sensor and
reference electrode which are most preferred comprises a
permselective layer, superimposed over at least a portion
of said sensor, having a thickness sufficient to exclude

30 substantially molecules with a molecular weight of about
120 or more while allowing the free permeation of molecules
with a molecular weight of about 50 or less; and a biolayer
superimposed over at least a portion of said permselective
layvr and said sensor, ' which biolayer comprises (i) a

35 sufficient amount of a bioactive molecule capable of

WO 92/01928 PCT/US91/05071

- 20 -

selectively interacting with a particular analyte species,
and (ii) a support matrix in which said bioactive molecule
is incorporated, which matrix is derived from the group
consisting of a photof ormable proteinaceous mixture, a
5 film-forming latex, and combinations thereof and through
which matrix said analyte species may freely permeate and
interact with said bioactive molecule.

In a preferred embodiment of the present
invention the permselective layer is derived from a polymer

10 film, most preferably comprising a heat-treated film of a
silane compound having the formula R'nSi (OR) 4-m in which n
is an integer selected from the group consisting of 0, 1,
and 2; R f is a hydrocarbon radical comprising 3-12 carbon
atoms; and R is a hydrogen radical or a lower alkyl radical

15 comprising 1-4 carbon atoms.

The bioactive molecule of the sensor biolayer
may be selected from a wide variety of molecules well known
to those skilled in the art and may include, for example,
an ionophore, an enzyme, a protein, polypeptide, nucleic
20 acid or an immunoreactive molecule. Typically, the
bioactive molecule is an ionophore or an enzyme.

In another preferred embodiment of the
microfabricated sensor, the photof ormable proteinaceous
mixture comprises (i) a proteinaceous substance; (ii) an

25 effective amount of a photosensitizer uniformly dispersed
in said proteinaceous substance; and (iii) water. Examples
of proteinaceous substances which are useful in the present
invention include albumin, casein, gamma-globulin,
collagen/ derivatives, and mixtures thereof. The most

30 preferred proteinaceous substance is an animal gelatin,
especially fish gelatin. Many types of photosensitizers
abound. Of particular interest, however, are high
oxidation state transition metal compounds, especially iron
and chromium salts.

WO 92/01928

PCI7US91/05071

- 21 -

The film-forming latex may comprise an aqueous
emulsion of a polymer or copolymer derived from synthetic
or natural sources.

Of course, additional layers may be present in
5 the microfabricated sensors. For example, additional
layers may be used to attenuate the transport of selected
molecules, including analyte species, through the sensor.
An electrolyte layer may be present especially for the
potentiometric sensors or the reference electrode
10 structure. A more complete description of the reference
electrode structure may be found in applicants' prior co-
pending U. S. Application Nos. 07/432,714 and 07/156,262.

It should be re-emphasized that an array of
sensors may use a common reference electrode. Thus, a

15 serieis of potentiometric sensors may be assembled for
measuring the activity of several electrolytes
concurrently, the signal of each sensor being determined
relative to the potential of the common reference
electrode. Amperometric sensors may have a slightly

20 different configuration, each comprising a sensor and a
counter electrode, for example, but with all the sensors in
an array sharing a common reference electrode.

The conductivity sensor for a hematocrit
measurement is plain in design comprising two noble metal

25 electrodes spaced at an appropriate distance on the
proposed array. In carrying out a conductivity
measurement, an electric field is generated between the
pair of metal electrodes by applying an a.c. signal (a d.c,
signal may also be employed). Preferably, the effect of

30 the field is limited predominantly to the fluid compartment
directly above the pair of metal electrodes. This
configuration maximizes the device's sensitivity toward
erythrocytes. It should be noted that several factors need

WO 92/01928

- 22 -

PCT/US91/05071

to be taken into account in selecting the appropriate
frequency for the a.c. signal. These factors include
minimizing Faradaic processes at the electrode surface
while maximizing the distortion of the field by the
5 erythrocytes .

The electrochemical sensors which are perhaps
the most complex, in terms of the need for additional
reagents, are those sensors used to perform kinetic
measurements. These types of sensors are useful in

10 determining, for instance, the activity of an enzyme as
reflected by the rate of change in the concentration of a
detectable species consumed or produced by the enzyme-
linked reaction. Hence, the activity of a particular
enzyme in a given sample may be established. Also, certain

15 enzyme-linked immunoassays may be carried out, paving the
way for the analysis of a wide variety of immunoreactive
and affinity-active species, including antigens, haptens,
antibodies, viruses and the like.

Thus, a preferred embodiment of a sensor
20 intended for enzyme or immunoassays should have a layer,
accessible to the sample fluid, to which is immobilized one
or the other of a ligand/ligand receptor pair. Again, the
reader is referred to the disclosure of applicants 1 prior
co-pending U. S. Application Serial No. 07/432,714 for
25 further details.

5.1.1. DISPOSABLE DEVICE FOR SENSORS

The microfabricated sensors described above are
preferably contained in a disposable device which can be
adapted for performing a variety of measurements on blood
30 or other fluids . The disposable device is constructed to
serve a multiplicity of functions including sample
collection and retention, sensor calibration and
measurement. During operation, the disposable device may

WO 92/01928

- 23 -

PCT/US91/05071

be inserted into a hand-held reader which provides the
electrical connections to the sensors and automatically
controls the measurement sequence without operator
intervention,

5 A suitable disposable device includes upper and

lower housing members in which are mounted a plurality of
sensors and electrical contacts and a pouch containing a
calibrant fluid. The sensors generate electric potentials
based on the concentration of specific ionic species in the
10 fluid sample tested. A double sided adhesive sheet is
situated between the upper and lower housing members to
bond the housing members together and to define and seal
several cavities and conduits in the device.

A first cavity is located at the center of the
15 device having a pin at the bottom of the cavity and a
hinged disc at the top of cavity. A sealed pouch
containing calibrant fluid resides in the cavity and a
first conduit leads from this cavity toward the sensors.
A second conduit has orifice at one end for the receipt of
20 a fluid sample while the other end of the tube terminates
at a capillary break. A third conduit leads from the
capillary break across the sensors * to a second cavity
which serves as a sink. The first conduit joins the third
conduit after the capillary break and before the sensors.
25 A third cavity functions as an air bladder. When the air
bladder is depressed, the air is forced down a fourth
conduit into the second conduit.

In operation, a fluid sample is drawn into the
second conduit by capillary action by putting the orifice
30 at one end of the conduit in contact with the sample.
After the sample fills the second conduit, the orifice is
sealed off. The pouch containing the calibrant fluid is
then pierced by depressing the disc down on the pouch which
causes the pin to pierce the other side of the pouch. Once

WO 92/01928

- 24 -

PCT/US91/05071

the pouch is pierced, the calibrant fluid flows from the
cavity through the first conduit to the third conduit and
across the sensors at which time the sensor calibration is
performed. Next, the air bladder is depressed forcing air
5 down the fourth conduit to one end of the second conduit
which forces the sample out the other end of the conduit,
past the capillary break, and into the third conduit and
across the sensors where measurements are performed. As
this is done, the calibration fluid is forced out the third
10 conduit into the second cavity where it is held. Once the
measurements are made r the disposable device, can be
discarded.

The hand-held reader includes an opening in
which the disposable device is received, and a series of

15 ramps which control the test sequence and the flow of the
fluid across the sensors. As the disposable device is
inserted into the reader, the reader ruptures the pouch of
calibrant fluid by depressing the hinged disc. The reader
then engages the electrical contacts on the disposable

20 device, calibrates the sensors, depresses the air bladder
to force the fluid sample across the sensors, records and
electric potentials produced by the sensors, calculates the
concentration of the chemical species tested and displays
the information for use in medical evaluation and

25 diagnosis.

Thus, for example, to measure the potassium
concentration of a patient's blood, the physician or
technician pricks the . patient f s finger to draw a small
amount of blood. The physician then puts the orifice of

30 the device into the blood, drawing the blood into the
device through capillary action. The physician then seals
off the orifice and inserts the device into the reader.
Upon insertion, a sequence of events is automatically
initiated by the reader without intervention from the

35 physician. The reader automatically causes the calibrant

WO 92/01928

PCT/US91/05071

- 25 -

pouch to be punctured so that the calibrant fluid flows
over the sensors, activating the sensors and providing the
necessary fluid for calibration. The electrical contacts
of the device are then automatically connected to the
5 reader and the calibration measurements are automatically
made. The reader then automatically depresses the air
bladder in the disposable device causing the sample to flow
over the sensors. The electric potentials generated by the
sensors are read and the concentration of the chemical
10 species is automatically calculated. The result is
displayed or output to a printer for the physician to
utilize.

Upon completion of the process, the physician
removes the device from the reader and disposes of it
15 properly. The reader is then ready to perform another
measurement which is initiated by the insertion of another
disposable device.

5.2. DATA HANDLING METHODS FOR PERFORMING
SIGNAL RESPONSE MEASUREMENTS

20 The present data handling methods allow the

instrument housing the external computational, data storage
and display means to extract the needed information (i.e.,
the electrochemical response of each sensor in the array)
from a background which includes sensor wet-up, fluidics

25 transients (those transients associated with fluid flow)
electronic noise, contact noise associated with the
electronic interface between the sensor and reference
electrode outputs and the computation means and other
intermittent artifacts or signal fluctuations. The reader

30 is referred to applicants 1 prior co-pending U. S.
Application Serial No. 07/187,665 for further information
concerning salient components of a handrheld instrument
which may be used for processing the sensors signals.

WO 92/01928

- 26 -

PCT/US91/05071

The data handling techniques include
computational methods which are designed to relate the
first and second signal measurement and, thus f provides the
concentration ratios of the preselected analyte species in
5 the first and second fluids. These computational methods
distinguish the relatively fast response of the sensor
(either potentiometric or amperometric) to concentration
changes from the slower monotonic wet-up of both the
sensors and reference electrode. After studying the

10 particular embodiments described herein, it should be
apparent to one skilled in the art, however, that more
sophisticated computational methods may be employed to
process signals, should non-monotonic wet-up behavior be
. encountered. Still other types of computational methods

15 may hereafter be conceived for detecting signal defects as
discussed below. In any event, a first order approximation
can be extended to a more general nth order polynomial
relationship, exponential relationship and the like, if the
need is apparent. The only concern with higher order

20 computations is their suitability for extrapolation.

Hence, in one embodiment of the present method a
computational method is employed to detect unusable signals
caused by: changes in the nature of the fluid in contact
with the sensor, electrical noise from the contacts or

25 connections between the individual sensors and the external
computational means, as well as other extraneous
intermittent artifacts. The present computational methods
may be used to provide an indication of the occurrence of
artifacts which combine to give an unacceptable measurement

30 cycle. In a preferred embodiment of the present method,
the standard computational method is extended to allow the
instrument to remove the offending artifacts or aberrant
data point from the acquired data set. The "corrected"
subset is then processed in the same way to provide a

35 useful measurement. Thus, certain but not all analysis,

WO 92/01928

- 27 -

PCT/US91/05071

which otherwise would have been di,. warded as "failures,"
are salvaged under appropriate conditions.

It should be pointed out that the computational
methods employed in the present method, though similar to
5 known signal processing methods, perform, inter alia, a
non-trivial assessment of how the electrochemical response
of interest and the wet-up behavior or artifacts contribute
to the raw waveforms. The desired signals must then be
derived or further manipulated before useful information is
10 obtained.

The present data handling method can be broken
down into two main parts comprising a data acquisition
portion and another for data manipulation and analysis.
Each portion has. its own set of computational methods, and
15 their relationship may be better understood by referring to
Chart I, below.

WO 92/01928

- 28 -

PCT/US91/0507!

CHART I

DATA ACQUISITION

1 SIGNAL CURRENTS 1 | SIGNAL VOLTAGE S?

ACTIVATION ALGORITHM
AND ELECTRONICS

ELECTRONICS
AND SOFTWARE

DATA MANIPULATION &HB ANALYSIS

| DATA MEMORY

UNACCEPTABLE DATA
NOT DISPLAYED

CORRUPT DATA
RECOVERY

SIGNATURE
ANALYSIS

ACCEPTABLE DATA
CALCULATION

DISPLAY ANSWER

WO 92/01928

- 29 -

PCI7US91/05071

During the data acquisition portion of the data
handling method, the analog signals obtained from the
sensors are converted into a digital format for recording
in the data memory. The electronics are designed to be
appropriate for high impedance potentiometric sensors with
sufficient resolution over the expected range of voltage
measurements. The electronics for the amperometric
sensors, which measure current, include current to voltage
converters and are also designed to have sufficient
resolution over the expected range of current measurements.
During the measurements, the fluid is grounded so as to
prevent the fluid potential from floating out of the range
of the operational amplifiers.

It should be pointed out that the amperometric
15 sensors are preferably subjected to an electrochemical
activation process . Applicants speculate that this
activation process enhances the catalytic activity of the
sensor electrode surface toward reduction or oxidation of
certain redox active chemical species. Further discussion
on this electrical activation process is presented in a
later section, below.

Also, this portion of the data handling method
includes the collection and digital storage of conductivity
measurements. As mentioned elsewhere in this disclosure,
25 these conductivity measurements are related to the analysis
of the patient's hematocrit levels and, also, system
quality assurance methods. It is important to note that
the wet-up of conductimetric sensors as described herein is
extremely fast (on the order of milliseconds) because the
sensor is comprised simply of two metal electrodes directly
in contact with the fluid. Hence, no extended wet-up of
intervening membranes is observed.

WO 92/01928

- 30 -

PCI7US91/05071

Additional aspects of the data acquisition,
manipulation and analysis are discussed in further detail
in a later section, below.

5.2.1. SIGNATURE ANALYSIS

5 The signature analysis section of Chart I is the

step in which the integrity of the signal measurements,
performed in the first and second data time windows, is
analyzed. The computational methods used in this section
detect the presence of excessive wet-up RC time constants,
10 spikes, glitches and noise and compare the values observed
within the time windows with preferred values held in
memory .

In a specific embodiment of the present method,
the first part of the signature analysis is run in real

15 time (i.e., during data collection) in which a seven point
sliding window slope analysis is implemented, beginning
with the first seven points of the time window. A total of
twenty-five data points (1-25) are actually collected in
each time window, whether the measurements are taken in the

20 presence of the first or second fluids. Although any size
can be chosen for the sliding windows, the seven point
sliding window provides an acceptable level of resolution.

The computational method is applied to each
sensor during data collection at each time window. In an

25 embodiment which utilizes an array of sensors, the
computational method keeps track of which sensor is active
and whether or not the time window has begun. The slope of
each seven point window is computed, based on a recursive
form of a linear regression. As each new point is

30 collected, causing the seven point window to slide, the
slope of the new seven point window (points 2-8) is
compared with the first seven point window (points 1-7) . A
range of acceptable values for the new slope based on the

WO 92/01928

- 31 -

PCT/US91/05071

value of the first slope (the basis slope) can be set, and
if any window is found to fall outside that range, a bit is
set for that sensor as a flag for later use. •

The present sliding window slope analysis
5 detects spikes and glitches in the time windows by looking
at first derivative changes. It should be apparent that
the slope of each new window can be compared with the slope
of the window which immediately preceded it instead of
comparing the new slope with that of the first seven point

10 window. The latter option saves time, however, by avoiding
the need to save in memory a new value for the basis slope
with each pass through the computational method. Also, the
present technique is more sensitive to low frequency
glitches than the alternate approach which involves a

15 trade-off in determining signal integrity. Yet another
alternative method could compare the slope of both nearest
neighbors rather than just the preceding point. Other
methods should be apparent to those skilled in the art.

The second part of the signature analysis

20 section involves a post-data collection processing
computational method that checks for limits on the observed
values. These computational methods may include the
calculation of first derivatives, error or estimates of
linear fit, delta drift rates, mean drift rates, second

25 derivatives, degree of curvature and the like. That is,
the computational method compares the observed data with
the expected range of values held in memory. In a specific
embodiment of the present method, limits are placed on the
drift rate (the slope) of each time window, the difference

30 between each time window's drift rate (i.e., the delta
drift rate between the first fluid time window and the
second fluid time window) and the mean or average value of
the response obtained in each time window. The drift rate
and mean values ar£ obtained from a linear regression

35 analysis. Maximum and minimum set values for each

WO 92/01928 PCT/US91/05071

- 32 -

parameter may be different for each sensor. Moreover, an
additional section of the post-data collection signature
analysis computational method computes the error associated
with the least squares fit and compares the value of this
error with the limit placed for each sensor in the array.
It is the selection of an optional combination of limits
that determines the accuracy and precision that is
attainable. For example tightening the noise limits but
relaxing the wet-up RC time constant may be preferable over
the reverse procedure.

Subsequent sections of the data manipulation and
analysis portion of the data handling package dictates
further actions if a particular sensors' time window(s)
contains spikes, glitches, noise or observed values (e.g.,
15 drift rates, delta drift rates, mean values or error of
estimates) falling outside an expected range. The results
of the affected sensors are not displayed and appropriate
advisories are then displayed over the instrument monitor.

However, if the data contain manageable

20 aberrations, a corrupt data recovery computational method
is then employed to derive sufficient information to
deliver a useful result. The corrupt data recovery
computational method includes a determination of whether
the detected glitch and/or spike is sufficiently large to

25 affect deleteriously the propriety of the linear fit
applied to the entire time window. This analysis is
accomplished by comparing the linear fit applied to the
entire time window to the basis slope obtained from the
first sliding seven point window. Alternatively, the

30 linear fits, with and without the offending glitch removed,
imay be compared and a decision made regarding its overall
integrity. Still another recovery computational method
implements a type of median filter to smooth detected
glitches. Again, other recovery methods may be

35 contemplated .

WO 92/01928

- 33 -

PCT/US91/05071

It may be useful, at this stage in the
discussion, to describe briefly the behavior of a
representative microfabricated sensor in the course of a
typical fluid analysis cycle. Figure 1 illustrates the
5 potential response, as a function of time, of a
potent iometric potassium ion sensor utilized in the present
invention. During the first several seconds, the
measurement cycle is initiated electronically, contact
being made between the array of sensors, which are mounted

10 preferably on a disposable assembly for example,

prior co-pending U.S. Application Serial No. 07/245,102,
the complete disclosure of which is incorporated herein by
reference), and the external computational means. During
this initial period, all the sensors are grounded. Within

15 a few seconds after the initial electrical contact is
established, the first fluid is caused to flow over and
make contact with the sensor array. As wet-up ensues the
potential of the potent iometric sensor drifts
monotonically, essentially in an exponential manner.

20 In the particular sequence illustrated in Figure

1, the fluid change is made after about 72 seconds, though
clearly, the introduction of the second fluid can be made
much sooner if obtaining this potassium ion concentration
is the only objective of the analytical cycle. In the

25 present illustration, the first time window can be selected
to fall at any suitable time after the first fluid is in
place over the sensors and prior to the introduction of the
second fluid. The second time window, in which the second
set of signal measurements is performed, is begun

30 preferably soon after the fluid change is made.

WO 92/01928

- 34 -

PCT/US91/05071

5.2.2, CALCULATION OF THE
SENSOR RESPONSE

Calculation of the sensor response, that is, the
change in potential reading between the first and second
5 fluid (the delta voltage) , is performed once an adequate
data set has been obtained. The present algorithm used to
calculate the sensor's response performs two linear least-
squares fits . One least-squares fit is performed in the
first fluid time window; the other is done in the second
10 fluid time window.

There exists many different ways to extract the
sensor response from a raw waveform such as that shown in
Figure 1. Some examples for deriving the delta response
value include, but are not limited to, a linear/linear

15 extrapolation, a linear/mean calculation, a mean/mean
approach or a mean/linear method, to name a few. In the
linear/linear case, the fit to the first time window is
extrapolated forward to an estimated fluid transition
point, the middle of the transition period. The fit to the

20 second time window is extrapolated backward to the same
estimated transition point, although clearly, it is not
necessary to specifically select the fluid transition point
at the midpoint . as some other suitable point at the may
also be selected) . The arithmetic difference between these

25 extrapolated voltages is the sensor's delta response.
Alternatively, the ratio between the extrapolated values is
calculated if currents derived from an amperometric sensor
are being measured. The concentration in the sample fluid
of the analyte species of interest can then be determined

30 based on the known chemical activity (concentration) of the
analyte in the calibrant fluid. The Nicolsky extension (1)
of the Nernst equation, which also takes into account the
effect on the sensor response of interfering ions present

WO 92/01928

- 35 -

PCT/US91/05071

in the fluids, is used for this determination, in the case
of potentiometric sensors:

E - E 0 + RT/nF log [A + k a ,h B] (1)

where E is the measured electromotive force (signal) , R is
5 the gas law constant, T is the absolute temperature, n is
the absolute value of the charge on analyte species a_
(e.g., n = 1 for the ammonium ion), F is the Faraday
constant, A is the activity of the analyte species a, B is
the activity of an interfering chemical species h, k a . h is

10 the interference coefficient associated with the effect of
the presence of chemical species h on the electrochemical
potentiometric determination of the activity of the analyte
species and E D is a constant independent of T, A or B.
For additional discussion of the Nicolsky equation, please

15 refer to Amman, D. Ion-Selective Mir roelectrodfis r Springer,
Berlin (1986) p. 68 and references cited therein, the
complete disclosures of which are incorporated herein by
reference.

A correction of the calculated chemical activity
20 of the analyte in the unknown or sample fluid can be
obtained by applying the Henderson equation (2) where the
term Ej is included in equation (1) and taking into
consideration differences in ionic strength and matrix
effects between the calibrant and sample fluids, which
25 differences are usually manifested as a slight response of
the reference electrode:

Xi 1 1 Zj I Hi/zi3 [ci (p) -c i (tt) ) RT ZilZiljiiCi ta)

Ej = — ; In — (2)

liUzilHi] [ci(p)-Ci(a)] F liUimiCifp)

where zi is the charge, \i± is the mobility, Ci is the molar
concentration of species i and a and p are transfer
30 coefficients.

WO 92/01928

- 36 -

PCT/US91/05071

For amperometric sensors , the concentration of
the analyte in the sample fluid is derived from the
measured current ratio r the known concentration of the
preselected analyte in the calibrant fluid and a system
5 constant. Equation 3 may thus be used:

[Ux/i2)ciY] + 5 « c 2 (3)

where ii and ±2 are the measured currents of the first and
second fluids and ci and C2 are molar concentrations of a
preselected analyte. If the first fluid is calibrant, then
10 ci is known and the value of C2 in the sample can be
readily obtained. The correction factors y and 8 are
derived experimentally and take into account differences in
the physicochemical properties of the calibrant and sample
fluids.

!5 m the linear/mean approach, the linear fit to

the calibrant time window is extrapolated forward, as
before, but to the midpoint of the sample time window. The
difference between the extrapolated value from the
calibrant fit and the mean sample value is the delta

20 response. Again, the ratio of the extrapolated value of
one time window to the midpoint value of the second time
window can be calculated also, if desired. The mean/linear
case reverses the direction of the extrapolation used in
the preceding approach, and the execution of the mean/mean

25 method of calculation should be fairly evident.

Ultimately, the method of choice for calculating
the sensor response depends on the characteristics of that
sensor and may be best determined through routine
experimentation. However, the quality of the sensor
30 response measurements will certainly have an impact on
which method is most appropriate. For instance, a frirly
large difference between the slopes of the data points in
the first and second time windows may indicate that a

WO 92/01928

- 37 -

PCT/US91/0507!

linear method may lead to a skewed result and that a mean
value of the particular time window may be more
appropriate. As already discussed, above, the
computational methods utilized during data collection,
5 manipulation and analysis may be enhanced to look at drift
rates (slopes) as data points are collected in each time
window to determine the "smoothness" of the window.
Likewise, first and second derivatives of fully sampled
time windows can be compared using "enhanced" computational
10 methods.

5.3. C A LI BRANT FLUID

The calibrant fluid should contain a known
concentration of the preselected analyte. The calibrant
fluid may comprise a wet gas but is preferably comprised of

15 an aqueous liquid. The chosen concentration of the
preselected analyte in the calibrant is preferably similar
to that expected to be encountered in the unknown sample.
When necessary, preservatives (e.g., p-hydroxybenzoate,
phenylmercuricacetate, p-aminobenzaldehyde and the like)

20 may also be included to prevent microbial contamination.
In the preferred method of the present invention, the
calibrant fluid is an aqueous solution of several analytes
whose concentrations are similar to those expected to be
determined in an unknown sample, usually whole blood/ It

25 has been the experience of the inventors that, due to
differences in the properties (e.g., viscosity) of the two
fluids, the practice of the present method is simplified if
a small volume* of blood- is used to displace a small volume
of calibrant fluid. That is, the first fluid is preferably

30 the calibrant,- and the second fluid is preferably the blood
sample. It should be evident, however, that the present
method is not so limited,, and the sequence and nature of
the fluids introduced to the sensors is a matter of choice.

WO 92/01928

- 38 -

PCT/US91/05071

5.4. POTENTIOMETRIC SIGNAL RESPONSE

Although the signal response phenomena of an
ion-selective electrode (ISE) upon change of the chemical
composition of the fluid with which the electrode makes
5 contact have been the object of much investigation in the
chemical literature, the time-dependent signal of a dry-
stored ISE upon first contact (wet-up) with an aqueous
medium has received relatively little attention. As stated
previously, this neglect results from the generally
10 accepted view in the art that prior to attaining a complete
equilibrated wet-up, the ISE can be of no analytical value.
However, while this premise may hold true for
conventionally fabricated macroelectrodes, it is not
necessarily true for microfabricated devices.

15 Clearly, to obtain useful measurements as close

to "real time" as possible, it would be very desirable to
record analytical readings from dry-stored ISEs before the
equilibrated wet-up process is complete. One means for
obtaining this measurement involves, the previously

20 described prior art differential method in which a pair of
electrodes with the same structure is used for each analyte
species and corresponding calibrant solution . . In such a
configuration, there is no conventional reference
electrode . The differential reading yields the ratio of

25 unknowns o-known concentrations with regard to that
particular chemical species to which that ISE is primarily
responsive. Based on the assumption that each of the pair
of " ISEs is identical in the characteristics that give rise
to the wet-up RC time-constant, e.g., physical dimension

30 and material composition, the expectation that the
monotonic wet-up signal will be canceled out of a
differential measurement appears to be reasonable.
Although an approach of some utility, it has certain
inherent limitations with regard to rapid, multi-species

35 electrochemical assays in parallel because it requires

WO 92/01928

- 39 -

PCT/US91/05071

Strictly simultaneous contact between each ISE of each
matched pair and the calibrant and sample fluids,
respectively. The problems associated with conducting and
using an array of these matched pairs of ISEs for the
5 analysis of multiple species would appear prohibitive as
these pairs have been used only for single analyte
determinations.

By the present method/ such limitations are
eliminated. The wet-up phenomena can be controlled to a

10 high degree of confidence, along with the physical
dimensions and material properties of the sensors and
reference electrode. In the present invention the output
shows that the wet-up has proceeded to a level where x < t
< 3x the slow rate of change can be treated as a linear

15 drift and this linear relationship can reliably project
the sensor's output forward in the time domain for some
brief period, e.g.,. 30 seconds. This capability becomes
crucial when one changes the solution in contact with the
sensor and reference electrode, e.g., (unknown) sample

20 (known) calibrant and compare the readings between the two
fluids. In this manner the need for strictly simultaneous
introduction of calibrant to the "reference" electrode and
the unknown sample to the "working" electrode is
circumvented.

25 It should be pointed out that even though a

mlcrof abdicated reference electrode is likely to have its
reference quality compromised by the eventual inward or
outward diffusion of ionic species from the aqueous medium,
this process is typically monotonic in the longer time

30 domain. As a consequence, output of a sensor measured
against such a reference electrode will be modeled reliably
by a linear relationship. Linear relationships are
generally preferred over higher order polynomials because
of the basic simplicity, less vulnerability of the signal

35 to interruptions, and greater confidence in obtaining a

WO 92/01928

- 40 -

PCT/US91/05071

meaningful extrapolation involved with the former.
Furthermore, it should be noted that the general
exponential time dependence of such relaxation phenomena
can be shown mathematically to be well approximated by a
5 linear function at sufficiently large values of time . The
propriety of this approximation becomes apparent if one
notes that the Taylor series expansion of the experimental
function is dominated by the linear terms as the argument
(-t/T) approaches (-a).

A series of Figures (1-3) follows to illustrate
the wet-up dynamics of several different wholly
microfabricated sensors with respect to a fully wet-up
conventional flow through silver-silver chloride reference
electrode, as well as the performance of such sensors
versus a microfabricated reference electrode that undergoes
similar wet -up effects. It can be seen that a linear fit
of the reading in one fluid enables one to predict
accurately what the value of the reading should be at later
times for comparison with a subsequent reading in a second
fluid.

•It is important to characterize the wet-up
behavior of a dry-stored, thin-film ion selective electrode
upon exposure to a calibrant fluid. As illustrated in
Figure 1, after the grounding path is eliminated and a
25 fluid path is established between the potassium sensor and
the reference electrode (at about 9 seconds interval) , the
raw waveform becomes more manageable. The exponential
decay of the potential output toward a steady-state value
is. a function of the wet-up of the potassium sensor and its
30 inherent RC time constant.

Figures 2a to 2e show the response
characteristics for a potassium, sodium, chloride, urea and
on-board microfabricated reference electrode, respectively,
versus a standard Corning reference electrode, with the

WO 92/01928

- 41 -

PCT/US91/05071

change from calibrant to the unknown (sample) fluid
occurring after about 50 seconds. (In these and subsequent
Figures the response of the sensor prior to introduction of
the calibrant fluid is not shown.) The wet-up response for
5 all of these ISEs is similar, showing a roughly linear
drift rate after about 40 seconds. It is particularly
important to note that the on-board microf abricated
reference electrode (Figure 2e) does not respond to the
fluid change, beyond minor correctable ionic strength and

10 matrix effects. Hence it may be used advantageously as the
actual reference electrode for the potassium, sodium,
chloride, and urea sensors. In Figures 3a-3d, the data
corresponding to these respective sensors in which the
microfabricated reference electrode is used as the on-board

15 reference demonstrates that, indeed, the on-board reference
electrode operates perfunctorily and that the glitches
associated with fluid change in Figures 2a-2d are even
eliminated. This result is observed due, presumably, to
closer proximity of the on-board reference electrodes to

20 the sensors.

In terms of the selection criteria for the type
of computational method for data manipulation and analysis,
an empirical approach can be employed which involves
determining the accuracy and precision for each of the

25 first order relationships. A Linear/Linear method is found
to be superior for the potassium, sodium and chloride
sensors. The urea sensor, which has a slower response-
time, provides best results when a Linear /Mean fit is used.
The exact location and duration of the data acquisition

30 windows can also be determined in this fashion,

5.5. AMPEROMETRIC SIGNAL RESPONSE

Unlike potentiometric measurements where the
ideal sensitivity of the response of a sensor is determined
by fundamental constants and intensive thermodynamic

WO 92/01928

42 -

PCT/US91/05071

quantities via the Nernst equation, the current output in
amperometric measurements is system dependent. That is,
the result is dependent upon the geometry and transport
properties of the overlaid structures, as well as the
5 surface properties of the electrode . However, over some
specific concentration range the current response will be
directly proportional to the bulk concentration of the
analyte. However the absolute current at any given bulk
concentration usually increases over the time domain, as

10 the dry-stored overlaid structures hydrate; that is, the
transport rate across the membranes increases. This
process can compromise the analytical value of the sensor,
limiting its achievable accuracy and precision, unless the
calibration process and subsequent measurement in a sample

15 (unknown) fluid are performed close together over the time
domain. Clearly, the operation of dry-stored, single-use
amperometric sensors require that the sequence of steps
performed over the course of the measurement be controlled
in a careful fashion.

20 In a preferred embodiment of the present

invention, at least two amperometric sensor signals (e.g.,
current output) are determined in each data aquisition time
window (again, there should be at least two time windows,
one for each fluid) . Most preferably, one of the sensor

25 signals determined in each time window (e.g., the calibrant
fluid time window) is measured at a first applied potential
and the other of the sensor signals, still in the same time
window, is measured at a second applied potential. A fluid
change is then made, and the above signal measuring process

30 is repeated.: If, for instance, the applied potential is
stepped up from one value to a higher applied potential in
the first time window, then in the second time window, it
may be convenient to step down from that higher value to a
lower applied potential.

WO 92/01928

- 43 -

PCT/US9J/0507J

The range of applied potentials at which the
individual signal measurements may be varied according to
the needs of the particular application. In the case of a
glucose sensor, such applied potentials may be chosen to
lie in the range of about 100 to about 300 mV. Most
preferably, the one of at least two signal measurements in
a given time window is carried out at an applied potential
of about 125 mV and the other measurement at about 225 mV.
The signal values obtained at each applied potential is
then plotted on a signal (e.g., current) versus applied
potential curve and the slope of the line defined by such
values calculated and compared for each time window. In
this manner, the slope of the line obtained for a calibrant
fluid having a known concentration of a preselected species
may then be compared with that recorded for a line derived
from a sample fluid.

As a further illustration of the present
embodiment, the attention of the reader is directed to
Figure 4 in which is shown an example of just such a signal
versus applied potential curve described immediately above.
On examination of the region of the curve lying between
about +100 to about +200 mV, one notes that a small net
positive current can be measured as the electrocatalytic
oxidation of hydrogen peroxide is not quite counterbalanced
by the corresponding reduction process. it is clear,
however, that in this region, or any given region between
about -100 to about +350 mV, the slope of the curve due to
plain HEPES buffer (X) is significantly different from that
of the curve due to a solution of 20 mM glucose in HEPES
buffer (0). In fact, the slope of each curve is directly
related to the concentration of glucose ' in each of the
different fluids.

Preferably, the applied voltages are centered
about the region in which the net current is or close to

WO 92/01928

- 44 -

PCT/US91/05071

zero. In this way, the contribution of interfering redox
processes to the total current or the effect of any bias
current associated with measuring electronics is minimized,
and the best estimate for the maximum slope is obtained.

5 5.6. SENSOR ACTIVATION

The mode of operation can be especially complex
for amperometric sensors which are active devices, unlike
potentiometric sensors which are electrically passive. An
amperometric sensor measures the rate of electron transfer
10 across the electrode-solution interface. Usually the
electrode surface plays a catalytic role in such electron
transfer (redox) reactions; therefore, the current is not
only dependent upon the surface area of the sensor but also
upon the catalytic activity of the surface.

15 It is not uncommon for electrode surfaces to be

contaminated or deactivated. While a catalytic iridium
surface, acting as the base sensor for a glucose electrode
( See r for example, prior U.S. Application Serial No.
07/432,714), is highly active towards hydrogen peroxide

20 oxidation prior to the deposition of overlaid structures,
its catalytic activity is much .reduced after such
processing. The following procedure is designed to recover
most of the sensitivity of the deactivated metal surface
without damaging the established overlaid architecture.

25 5.6.1. METHODS FOR GLUCOSE ACTIVATION

The present novel operational method is useful
for rapidly activating the electrode surface of a dry-
stored amperometric glucose sensor without deleteriously
affecting the overlaid structures . This activation makes
30 an important contribution to reducing the overall apparent
wet -up RC time-constant for the sensors. Unlike the prior
art pulsing methods, which are always carried out during
the analytical segment of the measurement (i.e., while the

WO 92/01928

- 45 -

PCT/US91/05071

devices are fully wet-up and the pulses are related to data
acquisition), the present activation is carried out in the
presence of a non-corrosive fluid but prior to data
acquisition and before the sensor has attained a fully wet-
5 up state.

To activate the structure, a set of pulse groups
is applied to the glucose sensor in the presence of the
calibrant or sample fluids. This process comprises
changing (cycling) the applied potential between values of

10 opposite sign (e.g., +1 V to -1 V). The pulses are applied
conveniently at full increment steps (i.e., at full ±l v
steps) . However, it should be apparent to those of
ordinary skill that such cycling may also be accomplished
in a variety of other ways including, but not limited to,

15 (i) pulsing; (ii) intermediate incremental steps to the
desired positive potential, followed by intermediate
incremental steps to the desired negative potential; (iii)
linear potential sweeps to the respective desired
potentials; and (iv) non-linear potential sweeps or such

20 sweeps which resemble smooth sinusoidal waves.

Broadly, the first two pulse groups in the set
have a magnitude and duration sufficient to activate the
catalytic surface of the electrode. These pulse groups
may also promote hydration of the overlaid structures . The
25 third pulse group in the set serves to reduce the initial
rate of change in current' upon application of the actual
operating potential. Referring now to Figure 5a the time
zero corresponds to the time at which the sensor comes into
contact with the fluid, calibrant in the preferred
30 instance), one observes that during the application of the
first two pulse groups, the peak currents for hydrogen
peroxide oxidation and reduction increase dramatically.
Figure 5b shows in more detail the response of the
activated sensor, at the operating potential, to a 5 mM
35 glucose calibrant solution (26-53 sec) followed by a 10 mM

WO 92/01928

- 46 -

PCT/US91/05071

human serum (from about 54 sec) . The analytical value of
the glucose sensor is clearly apparent, with a significant
current being observed for both fluids and in which the
current shows an increase after the sensor is exposed to a
5 biological fluid having a glucose concentration which is
higher than the calibrant solution. At an operating
potential of +350 mV, the sensor gives a linear current
response at a range of 0,1-30 mM glucose. Figure 5c,
suggests that the sensor is almost fully activated because
10 there is only a marginal increase in the maximum hydrogen
peroxide oxidation and reduction currents when the entire
pulsing sequence is repeated on the same sensor.

A viable alternative to cycling the applied
potential, involves galvanostatic control over the sensor
15 activation with a constant current source being applied to
the sensor until the required potential attains some
predetermined value or rate of change.

Another alternative for measuring the current
output of the sensors, to determine its level of
20 activation, involves making a conductivity measurement
between two electrodes, one which is the sensor.

In one embodiment of the present method the
pulsing sequence for hydrogen peroxide measuring sensors
are as follows : pulse groups 1 and 2 should be of a

25 duration of about 5 seconds with limits Of +0.7 to +1.2 V
at the oxidizing end and -0.7 to -1.2 V at the reducing
end, which pulses may or may not be applied symmetrically.
A single .extended negative pulse followed by the standard
sequence may also be applied advantageously. The inventors

30 have observed that negative pulses are an important aspect
of the activation process, and they speculate that the
activation process is likely to be associated with the
reduction of one or more types of functional groups of the
iridium electrode. However, the inventors do not wish to

WO 92/01928

- 47 -

PCT/US91/05071

be bound by the present speculation nor do they wish to
limit the scope of the present invention by making specific
interpretations of the surface activation process on a
molecular level.

5 The third P ulse group should be selected from

groups in which the final potential step is in the range
+400 mv to +800 mV. As can be seen from Figure 5b the
current recorded for the 5 mM glucose calibrant fluid shows
that the sensor has attained a sufficient degree of wet-up
10 and activation. During the period from 46 to 52 seconds,
data is recorded in the first time window for later
extrapolation. After 53 seconds the calibrant fluid is
forced to waste and the biological fluid (sample) placed
over the sensor. An advantage of thin-film microfabricated
biolayers is that the response-time (i.e., the time it
takes for the current to change to a value which is
proportional to the. bulk concentration of the analyte in
the second fluid) is fast, usually less than 5 seconds. It
is this additional property of the present microfabricated
devices that contributes to the success of the present
analytical methods. During the period from 61 to 66
seconds, the data is recorded in the second time window.

As with the potentiometric sensors, the
selection of an appropriate location for the data
25 acquisition window and the choice of the data fitting
computational method are based on empirical calculations of
the accuracy and precision attainable with each of the
fits. As with the potentiometric urea sensor a Linear/Mean
fit is preferred for the glucose sensor.

5.6.2. OPERATIONAL METHODS GENERALLY

The discussion above relates to a glucose sensor
comprising an iridium electrocatalyst at which hydrogen
peroxide produced by an enzymatic reaction is measured by

WO 92/01928

- 48 -

PCT/US91/05071

means of electrochemical oxidation at an operating
potential of +350 mV versus an on-board silver-silver
chloride reference electrode. An alternative method for
operating this sensor is based upon electrochemical
5 reduction of hydrogen peroxide. This operation is achieved
by applying an operating potential in the range of zero mV
to -250 mV, preceded by pulse groups similar to 1 and 2
above, but with pulse group 3 having a final potential set.
in the range of -200 mV to -500 mV.

10 The operational methods described above can also

be applied for activating sensors where platinum or another
noble metal is used as the electrocatalyst in place of
iridium; these surfaces also become deactivated during
post-processing steps*

15 An oxygen sensor of the type disclosed in the U.

S. Application Serial No. 07/432,714, with a gold
electrocatalyst suitable for oxygen reduction, may also be
operated with this type of activation method. However, it
is sometimes desirable to modify the duration and magnitude

20 of pulse groups 1 and 2 because different metal surfaces
are deactivated to different extents during the deposition
of overlaid structures and related processing steps. In
addition, pulse group 3 is preferably chosen to fall in the
range of -600 mV to -800 mV, where an operating potential

25 in the range of -400 mV to -550 mV is contemplated for
oxygen reduction at a gold electrode.

5.6.3. COMPUTATIONAL METHODS FOR

ENZYME ASSAYS AND IMMUNOASSAYS

In a metabolite assay the object of the
30 computational method is the determination of the bulk
concentration of the metabolite. Clearly, the sensor
should not perturb the bulk concentration of the metabolite
if the measurement is to be of analytical value.

WO 92/01928

- 49

PCT/US91/05071

For an enzyme sensor or an enzyme-linked
immunosensor-based assay, by contrast, the object of the
sensor computational method is the determination of the
rate of change in concentration of an electroactive species
which is consumed or produced while the assay is in
progress, which rate reflects the enzyme activity present
in the system. Sensors appropriate for enzyme and enzyme-
linked immunoassays are disclosed once again in U.S.
Application Serial No. 07/432,714.

If the product of the assay is hydrogen peroxide
(i.e., it is the electroactive species which is produced in
the course of the assay), the computational method can be
modified as follows: after a set of activation pulse groups
is applied and measurement of a calibrant current is
15 performed, as described above, substrates or reagents for
the enzyme or enzyme-linked immunoassay are introduced to
the sensor. After a brief induction or mixing period the
current changes steadily, and the initial rate of change,
di/dt, is then computed (Sea, Figure 6) . From the rate of
20 change in the current and the known current electroactive
species corresponding to a certain concentration in the
calibrant, the rate of increase in it's concentration can be
estimated. This rate can be expressed as the enzyme
activity, that is moles of substrate consumed per unit time
25 at a certain temperature, pH etc. It is important to note
that if the data collection period extends over several
minutes, the above-mentioned pulse groups wet-up the sensor
structure to a sufficient extent such that the calibrant
signal can be readily extrapolated forward in the time
30 domain. Otherwise the rate measurement, which is based on
the initial sensor response in the presence of calibrant
fluid (if calibrant is introduced prior to introducing the
sample), may be severely compromised (Refer to earlier
discussions related to the importance of making the
calibrant and sample measurements close in time) . The use

WO 92/01928

- 50 -

PCT/US91/05071

of microfabricated immunosensors in this manner makes it
possible to obtain rate measurements prior to the sensor
attaining a fully wet-up state.

Another problem that is encountered when a
5 sensor is operated over an extended period (2-3 minutes) is
that product build-up or reaction depletion zones occur in
the overlaid structures of the electrode with concomitant
diffusion gradients extending out into the bulk solution.
Such inhomogeneity close to the sensor can adversely affect

10 its performance. For example, an excessive differential
between the hydrogen ion concentration close to the
electrode and hydrogen ion concentration in the bulk
solution causes a shift in the activation energy for
hydrogen peroxide oxidation. This shift may result in a

15 non-linear current response. This problem may be
circumvented by modifying the computational method to
include additional sets of pulse groups at preselected time
intervals. After applying the initial set of three pulse
groups and the calibration has been performed, as described

20 above, substrates or reagents for the assay are allowed to
pass over the sensor* A measurement is then made in much
the same vay, as described above for glucose. However,
after this measurement the pulsing sequence, or some
modified portion thereof, is repeated and a second

25 measurement is made. This sequence is repeated several
times (usually five to ten) to yield a .set of current
measurements made at exact time intervals; The initial
rate of change in the current may then be more accurately
estimated from this set of current measurements because

30 repeated pulsing has the effect of washing-out the
electrode surface and overlaid structures and disordering
the fluid layer and concentration gradients which lie close
to the sensor. Again all of these processes are monitored
on a sensor that is undergoing wet -up, i.e., t ^ 3T.

WO 92/01928

- 51 -

PCT/US91/05071

5.6.4. ADVANCED OPERATIONAL METHODS

Advanced operational methods can also be used to
make the system "smart." Such methods can be applied
during the real time operation of the sensors or during
5 post data collection treatments. In particular, the
flexible methods for governing the activation of
amperometric sensors, assessing the propriety of a given
set of data points or determining the best position for the
data window may be incorporated, to name a few.

10 Thus, these methods may include alternatives to

applying the identical pulse sequence to all amperometric
sensors which may be present in the array. For example,
the operational method can be modified such that pulse
groups 1 and 2 are applied until the observed peak current

15 associated with each potential pulse has reached a finite
value, until the charge passed in each pulse has reached a
finite value, until the charge passed for the entire pulse
sequence has reached the finite value or until the RC time
constant for the sensors after a given pulse is within some

20 predetermined range of values. Moreover, pulse groups 1
and 2 may be applied until the rate of change between
successive pulses, in terms of peak current, charge passed
or RC time constant, is within some preselected range.

Advanced methods for error recovery can also be
25 implemented. In particular, computational methods can be
put in place for detecting the presence of glitches in the
data set, whether such glitches are isolated or numerous,
or whether they may be associated with common background
noise. An advanced method for sensing integrity failures
30 may suggest appropriate solutions, such as the application
of median filters. Alternatively, another data set may be
chosen which is derived from a separate data window stored
in the system. For example, if the sample causes an

WO 92/01928

- 52 -

PCT/US91/05071

abnormally slow response from the sensor, gives rise to
non-ideal waveforms upon contact with the sensors or,
generally, does not provide well-behaved waveforms, the
first and second time windows may not be positioned in the
5 best place initially. The presence of slight defects in
the the sensor may also, give rise to less than desirable
responses. In the case of a slow response, the maximum
value for the sample or second fluid may not have yet been
reached before a second signal measurement is performed

10 (i.e., a second time window is selected for comparison with
the first time window associated with the first or
calibrant fluid). The signal associated with the sample
may degrade rapidly soon after reaching the maximum value,
thus providing a lower value than what should have been

15 observed if the signal had not degraded.

In both cases discussed in the preceding
paragraph, the computational method may search for the
proper data collection window and locate the maximum value
whiclx provides a better estimate of the concentration of

20 the preselected analyte species. Also, even if certain
error limits are reached or exceeded, a reinspection of the
value of the slope and the quality of the overall drift may
prompt the instrument to accept the values obtained. In
certain cases, the value of the delta drift can override a

25 data window with an unusual number of glitches or a high
level of noise. Of course, different methods for
extrapolating the slopes and calculating the difference
between the signal measurements may be selected. These
non-ideal responses may be encountered more frequently with

30 sensors of increased functional complexity.

As mentioned earlier, one may also conceive of
flexible computational methods adapted to determining when
and whether a sufficient number of data points has been
collected. Such advanced methods may decrease the time

WO 92/01928

- 53 -

PCT/US91/05071

necessary for carrying out the data collection and
analysis.

5.7. CONDUCTIVITY MEASUREMENTS

For a colloidal suspension such as blood, the
electrical conductance is a function of the nature of the
particles, the medium and the fraction of the total volume
occupied by the particles. At high particle

concentrations, as in blood, it is also necessary to take
into account the geometry and orientation of the particle.

The entire period of data acquisition in the
calibrant and in the blood takes only about one second.
During this period the a.c. conductivity is measured at a
frequency which is selected to be sufficiently high to
minimize the impedance at the sensor-fluid interface, and
15 sufficiently low to minimize capacitive coupling across the
erythrocyte cell membrane. A preferred frequency is about
50 KHz .

Once the calibration has been made, the
calibrant fluid is then removed and blood introduced. The
measurement cycle is then repeated with the actual
measurement time being selected so that the blood is in a
quiescent state, but before a significant degree of
erythrocytes have settled.

The conductivity sensor comprises two noble
metal electrodes microfabricated on a planar surface and
designed with the appropriate geometry.

For conductivity measurements there is no need
for a sophisticated computational method for performing a
signature analysis. Calculation of the percent hematocrit
in the blood is made either from an empirically determined
calibration curve stored in the electronics or, more
preferably by means of an equation developed by Velick and

WO 92/01928

- 54 -

PCI7US91/05071

Gorin as described in Journal Ql General Physiology 1940,
21, 752-771.

Such conductivity measurement may also be
utilized to determine the success or failure of the fluid
5 change operation. That is , stored values of predicted
fluid conductivities can be compared to the observed
measurements to provide a quality assurance method for
detecting the presence of an adequate intervening air
segment between fluids, the failure of the calibrant fluid
10 to move to waste, or other such system failures associated
with the fluidics movements.

5.8. SYSTEM INTEGRATION

As alluded to elsewhere in this disclosure, a
system most attractive in the clinical setting is not

15 limited to discrete measurements of single analytes.
Instead, an array of sensors, designed to make a
multiplicity of discrete measurements of a range of
different analytes in biological fluids is preferred. This
array of sensors is preferably exposed to a single, common

20 calibrant. fluid which is removed after all of the sensors
have been calibrated. Only then can the second (sample)
fluid be introduced to the sensors. Such an integrated
setting means that all of the sensors must be chemically
and electrically compatible: that is, they must wet-up and

25 respond at approximately the same rate without interfering
with one another. Thus, sensor compatibility and system
integration is enhanced if the pulsing sequence, used in
amperometric measurements, is completed prior to
performing potentiometric measurements because the high

30 current flow in solution associated with pulsing can
undermine the integrity of the potentiometric signal.

Also, when the conductivity sensor is activated,
a current of ca. 10~ 3 A, flows in the solution between the

WO 92/01928

- 55 -

PCT/US91/05071

pair of metal electrodes. While this current may seem
small, it is about six orders of magnitude greater then
that measured in the amperometric based analyses. Again,
during the period when the amperometric sensors are being
5 pulsed, a maximum current of, ca. 10-6 A, also flows. To
minimize interference during calibration between the
conductimetric and amperometric sensors, it is preferable
that the conductivity sensor be operated after activation
of the amperometric sensors but before amperometric data
10 are acquired. After the fluid change, the amperometric
data acquisition is performed before the second
conductivity measurement is made. A series resistor, ca.
105 ohm, may also be used to protect amperometric sensors
during the a.c. conductivity measurements.

As mentioned previously, the conductivity sensor
may also be used to distinguish the general composition of
a fluid, i.e., whether the sensor is in contact with
calibrant fluid, serum or whole-blood, or for that matter
no sample at all, i.e., air. Because many of the sensors
display matrix effects, as discussed previously above, this
measurement may be used to make the appropriate correction
to the calculation of the concentration of the preselected
analytes.

In terms of the integration of conductivity
25 measurements with potent iometric measurements, there is no
discernable interference. However, it is preferred that
the conductivity measurement be avoided during data
acquisition at the potentiometric sensors.

In addition, it is preferred that there exists a
fluid grounding electrode in close proximity to the
potentiometric sensors to absorb excess charge generated in
the fluid when the conductivity sensor is activated. This
excess charge may have the undesirable effect of polarizing
the potentiometric membranes. Moreover, the membranes must

WO 92/01928

- 56 -

PCT/US91/05071

be depolarized rapidly if they are to be of analytical
value.

Finally, it may be desirable to place the
conductivity measuring sensor on a separate chip from that
5 of the potent iometric sensors. Commonly, these sensor
structures are actually fabricated on a silicon wafer with
an insulating layer, ca. 0.5 1CT 6 m in thickness, of
silicon dioxide . The underlying silicon is a semiconductor
which means that capacitive coupling between sensors
10 presents a possible signal interference. Separating the
sensors on different chips obviates this potential problem.

6 . EXAMPLES

As a further illustration of the present method,
the following example is described. Referring now to Chart

15 II, there is shown a more detailed flow chart of the Data
Collection method.; As mentioned previously, data
collection may be accomplished in several different ways,
including simply sequentially measuring and storing data.
The method of the preferred embodiment permits verifying

20 integrity of the measurements in real time, while
collecting input data. Although a thorough data integrity
analysis is provided at the Data Analysis stage, this on-
line verification method permits early detection and,
possibly, elimination of spikes, glitches, and other noise

25 in the data. Alternatively, time windows can be determined
intelligently at flexible intervals where the data points
are not significantly affected by noise. Such smooth time
windows can thus be located readily.

- 57 -

PCI7US9I/05071

CHART II

" \ COLLECT DATA POINT | —

22 [

INCREMENT DATA COUNTER

DATA COUNTER
£ SLIDING POINTS

COMPUTE RATIO 07
SLIDING WINDOW SLOPE
TO BASIS WINDOW SLOPE

SET

ERROR BIT IF

(RATIO

> UPPER BOUND)

AND

(RATIO

< LOWER BOUND)

COMPUTE

BASIS

WINDOW

SLOPE

WO 92/01928

- 58 -

PCT/US91/05071

In the present example, the Data Collection
process begins at block 21 , where each data point is
measured. Data acquisition involves taking an analog
signal and recording it in digital format in the
5 appropriate memory location. To ensure that the system is
not susceptible to 60 cycle noise, a set of 80 data points
every one-sixtieth of a second are preferably recorded.
(Clearly in countries wh£ re 50 cycle noise is the norm the
sampling rate is set at 80 data points every one-fiftieth
10 of a second) . The average value of this set is then
calculated, and this value is taken as a "single" data
point. These "single" data points are then used in further
calculations.

As described above, the electronics are designed
15 to be appropriate for high impedance potentiometric sensors
with sufficient resolution over the expected range of
measurements. The' electronics for amperometric sensors
includes current to voltage convertors designed to have
sufficient resolution over the expected range of current
20 measurements.

Next, the method flow passes to block 22, where
a counter (DATA_COUNTER) , which keeps track of the
aggregate number of data points collected within a time
window, is incremented by one, (The DATA_COUNTER is

25 initialized to zero prior to acquisition of the first data
point of a given time window.) The method flow then passes
to test 23, where the total number of collected data points
(DATA COUNTER) is compared to the number of data points of
a selected sliding window set (SLIDING_POINTS) . The size

30 of a sliding window is selected on the basis of a trade-off
between the computational time and precision of error
detection. In this instance, a seven point sliding vindow
is selected from a twenty-five point time window.
According to test 23, if the total number of acquired

WO 92/01928

- 59 -

PCT/US91/05071

points, is less than the size of one sliding window, the
control returns to block 21, where another data point is
acquired. Otherwise, control is transferred to test 24.

Test 24 is provided to locate the basis window.
5 In this instance, the basis window is defined as the first
sliding window, and remains the same throughout the data
acquisition process for a given time window. it should be
apparent to one of ordinary skill in the art that there are
other methods of selecting a basis window. For example, a
10 basis window can be defined as a sliding window which is
"behind" the current sliding window by one or more points.

According to test 24, if an exact number of
SLIDING_POINTS are collected, method flow is transferred to
block 29; otherwise, the total number of collected points

15 is greater than the size of one sliding window, and method
flow passes to block 25. At block 29, the slope of the
basis window is computed using a linear regression (i.e.,
by fitting points into a line defined by y = ax + b,
wherein the slope is "a"). At block 25, a new sliding

20 window set is formed and a recursive form of the linear
regression method is used to compute the slope of a sliding
window. m this instance, a new sliding window set is
created whenever flow enters block 25. Therefore, a first
sliding window consists of the following data points: {2, 3
25 ... (SLIDING_POINTS + 1)), and each subsequent sliding
window is formed by including a newly acquired data point
and eliminating the first data point of the previous
sliding window set.

Next, the method flow passes to block 26, where
30 the slope of the basis window is compared to the slope of
the current sliding window, and the ratio of the slopes is
computed. Then at block 27, the integrity of a current
sliding window data set is verified. If the ratio of the
slopes is not within acceptable bounds, the output of such

WO 92/01928

- 60 -

PCT/US91/05071

sensor is deemed to contain an error. One can set, for
example, an upper bound of 2.5 for the ratio. If, however,
the basis window has a value of zero, a difference instead
of a ratio is used.

5 Hence, if the ratio or difference is out of

bounds, a flag (ERROR) is set. In this instance, the ERROR
flag is simply a bit. Alternatively, the ERROR flag value
might contain a pointer to a specific data point that
causes the ratio of the slopes to be out of bounds. Next,

10 the method flow enters test 28, which checks whether
additional data points should be measured for a given time
window. If all the time window points <WJPOINTS) have not
been collected, the flow returns to block 41; otherwise,
the control is transferred to a new process step which

15 causes the displacement of the calibrant fluid by the
sample fluid, with an air segment present between the
fluids. When the last time window has been collected, the
following data analysis stage is carried out. It should be
apparent that W_POINTS are collected for each active sensor

20 in the array. Thus, each process block of Chart II is
performed essentially in parallel for each active sensor.

" Chart III shows schematically the steps of the
Data Analysis stage of the method in which analyte
concentrations are determined from each sensor measurement.
25 At this stage, data points for the first (e.g., calibrant)
and second (e.g., sample) time windows have been collected
and stored in appropriate memory locations. Test blocks
41-47 occur for each active sensor in the array, while test
blocks 48-51 occur once for the entire array.

30 The process begins at test 41 which checks

whether an error has been detected at the Data Acquisition
stage by testing the status of the ERROR bit. If the ERROR
bit is set, method flow passes directly to block 42, where

WO 92/01928

- 61 -

PCT/US91/05071

error recovery is performed. Otherwise (i.e., no errors at
Data Acquisition stage), flow passes to block 42.

- 62 -

PCT/US91/05071

CHART III

DETERMINE SLOPES IN
THE TIME WINDOWS

NO ERRORS

EXTRACT DELTA
RESPONSE

MARK SENSOR
FAILED

ERROR RECOVERY

ADJUSTMENTS DUE TO
REF ELECTRODE RESPONSE

CALCULATE
CONCENTRATION

STANDARDIZE

DISPLAY CONCENTRATIONS
OR ERROR MESSAGES

WO 92/01928

- 63 -

PCT/US91/0507!

At block 42, a linear regression is applied to
the data points of each window to approximate the drift
rates (slopes) of the data window. A linear least squares
fit is a well known method of interpolation, in which data
5 points are connected to approximate a line, w y = ax + b, "
where constants n a n and "b" are chosen such that the sum of
squares of the deviation from the actual data points is
minimized. Alternatively, higher order regressions, i.e.,
those that approximate 2nd, 3rd, nth order functions,

10 can be implemented easily. Generally, the order of the
applied regression should be determined according to the
nature of the sensors' degree of wet-up and waveform in the
selected time window intervals.

Next, the flow passes to test 43, where the data
15 points, of both windows are checked for unexpected values.
The following data integrity verification is then
performed:

MIN_SLOPE < first window slope < MAX_SLOPE •
MIN_SL0PE < second window slope < MAX_SLOPE
20 (first window slope - second window slope) < MAX_DIFF

MIN_MEAN < first window mean -value < MAX_MEAN
MIN_MEAN < second window mean value < MAX_MEAN_1
first window error of approximation < MAX_ERR0R
second window error of approximation < MAX_ERROR

25 The values, MIN_SL0PE, MAX_SLOPE, MAX_DIFF,

MIN__MEAN, MAX_MEAN_1 and MAX_ERROR are expected ranges of
the above parameters. The specific values are determined
experimentally, based on the predictable characteristics of
the sensors' response curve. Table I lists some suggested

30 values for the various data window parameters.

WO 92/01928

- 64 -

PCT/US91/05071

TABLE I

SUGGESTED LIMITS FOR VARIOUS DATA WINDOW PARAMETERS
. FIRST WINDOW

SLOPE

(mV/s)

ME AM

(mV)

ERROR (mV)

min

max

min

max

-0.65

0.65

-200

+100

0.2

-0.25

1.0

-250

+100

0.2

0.0

1.0

-250

+100

0.2

BUN a

-0.65

0.65

-200

+100

0.2

Glue 15

-0.02 c

0.02 c

0.2 d

0.015 d

SECOND

.WINDOW

SLOPE

(mWs)

mean,

, (mV).

ERROR (mY>

SENSOR

min

max

min

max

max.

-.0.65

0.65

-200

+100

0.2

-0.25

1.0

-250

+100

0.2

0.0

1.0

-250

+100

0.2

BUN a

-1.0

1.0

-200

+100

0.4

Gluc*>

-0.1 c

0.1 c

0.05 d

32 d

0.015 d

20 a CI, K, Na and BUN stands for chloride, potassium, sodium
and Blood Urea Nitrogen sensors, respectively. b Gluc
stands for Glucose sensor. c This value is in units of
nA/s. d This value is in nA.

WO 92/01928

- 65 -

PCT/US91/0507I

In the present instance, the limit on the
difference (in mV/s) between the value of the slope in each
time window is set at a maximum of 0.50, 0.30, 0.30 and
1.00 for the chloride, potassium, sodium and BUN sensors,
5 respectively. The observed waveform for the second fluid,
such as glucose, in contact with the microf abricated
amperometric sensors can exhibit a maximum or minimum'
value; consequently a linear/mean fit is preferred.

If one of the above tests indicates that the
10 measurement is out of range, all the measurements generated
by a particular sensor are discarded as unreliable.
However, the present method provides for error recovery of
corrupt data. Accordingly, if an error is detected at test
43, the method flow passes to test 44. If no errors are
15 detected, control is transferred to block 47, where the
sensors 1 responses are determined.

Test 44 determines whether a previous attempt to
correct errors has already been made. At this step, a
variable "FIRSTJTIME" is incremented by one. (FIRST_TIME

20 is initialized to zero at the beginning of the Data
Analysis for each time window.) A value of FIRSTJTIME that
is greater than 1 indicates that a previous attempt to
correct errors has been made. Because the flow returned
subsequently to test 44, error recovery failed once and

25 should not be repeated. In this case, method flow passes
to block 46, where the "FAILED" bit is set to indicate that
this particular measurement has failed, and that the only
further processing remaining for such sensor involves
displaying an error message at block 51. Otherwise, the

30 value of FIRST_TIME is one and the flow passes to block 46,
where error recovery is attempted.

Error recovery can be implemented in a variety
of ways. For example, noise and spikes in the data can be

WO 92/01928

- 66 -

PCT/US91/05071

eliminated by known digital techniques, such as median
filters and the like described previously, i.e., the re-
evaluation of observed values e.g., p. 29.) Another
method may involve the interpolation of a curve on the
5 basis of the first several points of a given time window
and the rejection of those points that deviate
significantly from .the interpolation. In most cases, it
may be desirable to collect more data points than that
required for a given time window, so that error recovery
10 may be accomplished by selecting an alternative time window
set. From error recovery at block 4 6, method flow turns
back to block 42 where the points of the corrected data set
are interpolated and new slopes and errors of approximation
are computed.

15 if collected data passes all the integrity

tests, the control is transferred to block 47 where the
sensors 1 responses are determined by relating measurements
in the first (calibrant) time window to those in the second
(sample) time window. For potent iometric sensors, the

20 analytical value of interest is the delta response, which
is the difference between the amplitudes measured at a
selected point of an analyte and calibrant response curves .
For amperometric sensors, the delta response corresponds to
the ratio of the respective amplitudes • As mentioned

25 above, the data acquisition is performed while the sensors
exhibit a monotonic wet-up response. In this instance, the
responses are measured in time windows which are selected
such that linear interpolations are sufficiently accurate
to describe the response drifts. Consequently, linear

30 approximations of block 42 can be reliably projected
forward and backward in time to compare each sensor's
response to the fluid change, AS mentioned previously, nth
order approximations can also be used for this purpose.

The particular sensor's delta response is then
35 calculated in one of the many different ways including, but

WO 92/01928

67 -

PCI7US91/05071

not limited to, a Linear/Linear delta, Linear/Mean delta,
Mean/Mean delta, Mean/Linear delta, Maximum/Linear delta,
Linear/Maximum delta, Linear /Minimum delta and
Minimum/Linear delta approaches. In the Linear/Linear
5 case, the fit to the first window is extrapolated forward
to an estimated fluid transition point, i.e., the midpoint
between the first window and the second window. The fit to
the second window (sample) is extrapolated backward to the
same estimated transition point. The difference between
10 these extrapolations is the particular sensor's delta
response. In the Linear/Mean approach, the linear fit to
the calibrant is extrapolated forward to the midpoint value
of the sample window and compared to the midpoint value of
the sample response curve. In the Mean/Linear approach the
15 previous sequence is reversed. In the Mean/Mean approach,
the delta response is the difference between the midpoint
of the sample window and the midpoint value of the
calibrant window. ; The Linear /Maximum, Maximum/Linear,
Linear/Minimum and Minimum/Linear methods are analogous to
20 Linear /Mean and Mean/Linear methods, except that the mean
value is replaced by the maximum or minimum value of the
corresponding time window. On the basis of routine
experimentation, one of the above methods can be found more
desirable than the others for obtaining a more accurate
25 measure of the concentration of a particular analyte when
compared to a reference method of analysis. Presently, the
Linear/Linear method is found to be superior for the
potassium, sodium and chloride sensors. A Linear /Mean
method is best for the present embodiment of the glucose
30 and urea sensors.

^ After the delta responses are computed for each
sensor, the flow passes to block 48. At block 48, the
response of the reference electrode is subtracted from the
delta response determined in block 47. At this stage,
35 certain corrections are made for a slight response of the

WO 92/01928

- 68 -

PCT/US91/05071

reference electrode due to matrix effects and differences
in ionic strength of the fluids. Next, the flow passes to
block 49, where, for potentiometric sensors, Equation 1 and
coefficients derived from Equation 2, as described above,
5 are applied to determine analyte concentrations. For
amperometric sensors, Equation 3 is applied. Then, at
block 50, the results are appropriately scaled in order to
derive standardized values. Finally, block 51 provides a
display of the calculated concentrations or a display of an
10 appropriate message if aberrant values are found.

The preceding example is presented solely to
illustrate a method for practicing the invention and should
not be construed as limiting the invention in any way.
Doubtless, other embodiments may be conceived which would
15 not depart significantly from the spirit and of the present
invention, which scope is defined by the following claims.

WO 92/01928

69 -

PCT/US9I/05071

WHAT IS CLAIMED IS:

1. A method of determining the concentration
of a preselected analyte species present in a sample fluid
comprising:

(a) providing an external computational means,
a reference electrode and at least one substantially dry-
Stored sensor capable of exhibiting a response to changes,
in the concentration of a preselected analyte species
before said sensor attains full equilibrated wet-up;

(b) establishing electrical contact between
said sensor, reference electrode and external computational
means; .

15 < d > performing a first signal measurement in a

first time window in the presence of said calibrant fluid;

(e) displacing said calibrant fluid after
performing said first signal measurement;

(f) contacting said sensor and reference
20 electrode with a sample fluid;

(g) performing a second signal measurement in a
second time window in the presence of said sample fluid;
and

(h) relating said first and second signal
25 measurements to determine the concentration of said

preselected analyte species in said sample fluid before
said sensor attains full equilibrated wet -up.

WO 92/01928

- 70 -

PCT/US91/05071

2. A method of determining the concentration
of a preselected analyte species present in a sample fluid
comprising:

(a) providing at least one microf abricated
5 potentiometric sensor sensitive to changes in the
concentration of a preselected analyte species , a reference
electrode capable of sustaining a well-behaved reference
potential for a sufficient period of time and external
computational means,

!0 which sensor and reference electrode have been

stored substantially dry and which sensor exhibits a rapid
response to said changes in the concentration of said
preselected analyte species before said sensor attains full
equilibrated wet-up;

!5 (b) establishing electrical contact between

said sensor, reference electrode and external computational
means;

20 (d) performing a first signal measurement in a

first time window in the presence of said calibrant fluid;

(e) displacing said calibrant fluid after
performing said first signal measurement;

(f) contacting said sensor and said reference
25 electrode with a sample fluid;

(g) performing a second signal measurement in a
second time window in the presence of said sample fluid;
and

WO 92/01928

- 71 -

PCT/US91/05071

.(h) relating said first and second signal
measurements to determine the concentration of said
preselected analyte species in said sample fluid before
said sensor attains full equilibrated wet-up.

5 3. A method of determining the concentration

of a plurality of different preselected analyte species
present in a sample fluid comprising:

(a) providing an array of raicrof abricated
potent iometric and amperometric sensors, each sensitive to

10 changes in the concentration of different preselected
analyte species, and external computational means,

which array comprises individual sensors and at
least two reference electrodes capable of sustaining a
well-behaved reference potential for a sufficient period of

15 time, one reference electrode being connected to said
potentiometric sensors when the other reference electrode
is connected to said amperometric sensors, and each of
which sensors has been stored substantially dry and
exhibits a response to said changes in the concentration of

20 the different preselected analyte species before each
attains full equilibrated wet -up;

(b) establishing electrical contact between
said array of sensors and external computational means;

(d) performing a first set of signal
measurements in a first time window in the presence of said
calibrant fluid;

(e) displacing said calibrant fluid after
30 performing said first set of signal measurements;

WO 92/01928

- 72 -

PCIYUS91/05071

(f) contacting said array of sensors with a
sample fluid;

(g) performing a second set of signal
measurements in a second time window in the presence of

5 said sample fluid; and

(h) relating said first and second sets of
signal measurements to determine the concentration of a
plurality of the different preselected analyte species in
said sample fluid before each sensor attains full

10 equilibrated wet-up.

4 . A method of determining the ratio of the
concentrations of a preselected analyte species present in
two fluids comprising:

(a) providing at least one microf abricated
15 amperometric sensor sensitive to changes in the
concentration of a preselected analyte species, a reference
electrode capable of sustaining a well-behaved reference
potential for a sufficient period of time and external
computational means,

20 which sensor and reference electrode have been

stored substantially dry and which sensor exhibits a rapid
response to said changes in the concentration of said
preselected analyte species before said sensor attains full
equilibrated wet-up;

25 (b) establishing electrical contact between

said sensor, reference electrode and external computational
means;

WO 92/01928

- 73 -

PCT/US91/05071

(d) performing a first signal measurement in a
first time window in the presence of said first fluid;

(e) displacing said first fluid after
performing said first signal measurement;

(f) contacting said sensor and said reference
electrode with a second fluid;

(g) performing a second signal measurement in a
second time window in the presence of said second fluid;
and

10 (h) relating said first and second signal

measurements to determine the ratio of the concentrations
of said preselected analyte species in said first and
second fluids before said sensor attains full equilibrated
wet -up.

15 5 • A method of determining the ratio of the

concentrations of a preselected analyte species present in
two fluids comprising:

(a) providing at least one dry-stored
microfabricated sensor capable of exhibiting a response to
changes in the concentration of a preselected analyte
species, which response is fast relative to the monotonia
wet-up behavior of said sensor when contacted with a fluid;

(b) providing a dry-stored microfabricated
reference electrode capable of sustaining a well behaved

25 reference potential for a period of time sufficient to
permit the completion of at least two signal measurements
and exhibiting a monotonic wet-up behavior which is similar
to that exhibited by said sensor when contacted with a
fluid;

WO 92/01928

- 74 -

PCT/US91/05071

(d) contacting said sensor and reference
5 electrode with a first fluid;

(e) performing a first signal measurement in a
preselected first time window in the presence of said first
fluid before said sensor and reference electrode attain
full equilibrated wet-up;

10 (f) displacing said first fluid after

performing said first signal measurement;

(g) contacting said sensor and reference
electrode with a second fluid;

(h) performing a second signal measurement in a
15 preselected second time window in the presence of said

second fluid before said sensor and reference electrode
attain full equilibrated wet-up;

" (i) relating the first and second signal
measurements to determine the ratio of the concentrations
20 of said analyte species in said first and second fluids by
a computational method which distinguishes the relatively
fast response of the sensor to changes in the concentration
of said preselected analyte from the slower monotonic wet-
-up behavior of said sensor and reference electrode.

6. The method of claim 1, 2 or 3 in which the
order by which said sensor, reference electrode or array
thereof is contacted with said calibrant and sample fluids
is reversed.

7. The method of claim 6 in which said first
30 and second signal measurements or sets thereof are

WO 92/01928

- 75 -

PCT/US91/05071

performed in the presence of said sample and calibrant
fluids, respectively.

8. The method of claim 4 or 5 in which said
first fluid is a calibrant fluid.

5 9 - The method of claim 8 in which said second

fluid is sample fluid.

10. The method of claim 4 or 5 in which said
first fluid is sample fluid.

11. The method of claim 10 in which said second
10 fluid is calibrant.

12. The method of claim l, 2, 3, 4 or 5 which
further comprises providing a conductivity sensor capable
of measuring the conductivity of a fluid in contact
therewith .

15 13 • The method of claim 1, 2 or 3 which further

comprises performing a first conductivity measurement in
the presence of said calibrant fluid.

14. The method of claim 4 or 5 which further
comprises performing a first conductivity measurement in

20 the presence of said first fluid.

15. The method of claim 13 which further
comprises performing a second conductivity measurement in
the presence of said sample fluid.

16. The method of claim 14 which further
25 comprises performing a second conductivity measurement in

the presence of said second fluid.

17. The method of claim 13 in which said first
conductivity measurement is performed prior to performing
said first signal measurement or set thereof.

WO 92/01928

- 76 -

PCI7US91/05071

18. The method of claim 14 in which said first
conductivity measurement is performed prior to performing
said first signal measurement.

19. The method of claim 15 in which said second
conductivity measurement is performed after performing said
second signal measurement or set thereof.

20. The method of claim 16 in which said second
conductivity measurement is performed after performing said
second signal measurement.

21. The method of claim 3 or 4 which further
comprises activating said amperometric sensor or array
thereof .

22. The method of claim 21 in which activating
said amperometric _ sensor comprises subjecting said

15 amperometric sensor 'to a series of potential changes in the
presence of said calibrant or first fluid.

23. The method of claim 22 in which said
potential. changes comprises cycling the applied potential
between values of opposite sign relative to said reference

20 electrode.

24. The method of claim 22 in which said series
of potential changes includes a first group which comprises
at least ten repetitions of a cycle in which the applied
potential is maintained at a first value and then is

25 switched, to a second value of equal magnitude but of
opposite sign as said first value.

25. The method of claim 24 in which said series
of potential changes further includes a second group which
comprises at least five repetitions of a cycle in which the

30 applied potential is maintained at a first value and then

W ° 92/01928 PCT/US91/05071

- 77 -

is switched to a second value of equal magnitude but of
opposite sign as said first value.

26. The method of claim 25 in which said series
of potential changes further includes a third group which
comprises at least five repetitions of a cycle in which the
applied potential is maintained at a first value and then
is switched to a second value of equal magnitude but of
opposite sign as said first value.

27. The method of claim 24 in which the
application of said first group is commenced within about
one second after said calibrant or first fluid comes into
contact with said amperometric sensor.

28. The method of claim 25 in which the
application of said second group is commenced within about

15 twelve seconds after said calibrant or first fluid comes
into contact with said amperometric sensor.

29. The method of claim 2 6 in which the
application of said third group is commenced within about
twenty seconds after said calibrant or first fluid comes
into contact with said amperometric sensor.

30. The method of claim 25 in which said first
value for said first and second groups is about +1000 mV.

31. The method of claim 26 in which said first
value for said third group is within double the magnitude

25 of the operating potential of said amperometric sensor.

32. The method of claim 3 which further

comprises :

providing a conductivity sensor capable of
measuring the conductivity of a fluid in contact therewith,

WO 92/01928

- 78 -

PCT/US91/05071

activating said amperometric sensor in the
presence of said calibrant fluid, and

performing a first conductivity measurement in
the presence of said calibrant fluid, after activating said
5 amperometric sensor but before performing any set of signal
measurements .

33. The method of claim 32 which further
comprises performing a second conductivity measurement in
the presence of said sample fluid, after performing all

10 sets of signal measurements.

34. " The method of claim 1, 2 f 3, 4 or 5 in
which said signal measurements or sets thereof are
performed within about 2 minutes

35. The method of claim 1, 2, 3, 4 or 5 in
15 which the end of said first time window and the beginning

of said second time window are separated by about three to
about six seconds.

.36. The method of claim 1, 2, 3, 4 or 5 in
which the end of said first time window and the beginning
20 of said second time window are separated by no more than
about ten seconds.

37. The method Of claim 1, 2, 3, 4 or 5 in
which the duration of said time windows is about five to
about fifteen seconds.

25 38. The method of claim 3 in which one of said

reference electrodes is replaced with a counter electrode,

such that the remaining reference electrode is
available to both of said potentiometric and amperometric
sensors and said counter electrode is dedicated to said
30 amperometric sensors •

WO 92/01928 PCT/US91/05071

- 79 -

39. The method of claim 1, 2, or 3 in which the
displacement of said calibrant fluid is carried out by the
introduction of said sample fluid.

40. The method of claim 39 in which the
5 displacement of said calibrant fluid with said sample fluid

is carried out with an air segment separating said fluids.

41. The method of claim 4 or 5 in which the
displacement of said first fluid is carried out by the
introduction of said second fluid.

10 42 ' The method of claim 41 in which the

displacement of said first fluid with said second fluid is
carried out with an air segment separating said fluids.

43. The method of claim 1 in which performing
said signal measurements comprises acquiring a preselected

15 number of data points over the time period of said time
windows .

44. The method of claim 43 in which acquiring
said data points comprises:

(a) collecting a first preselected fraction of
said data points in a first sliding window, which first
sliding window comprises a corresponding fraction of said
time window; and

(b) collecting additional preselected fractions
of said data points in subsequent sliding windows until

2.5 every data point in said time window is included in at
least one sliding window.

45. The method of claim 44 in which said first
subsequent sliding windows are of equal duration.

WO 92/01928

- 80 -

PCT/US91/05071

46. The method of claim 44 in which said first
and additional preselected fractions contain an equal
number of data points .

47. The method of claim 44 in which collecting
5 additional fractions of said data points comprises

including at least one new point of data in said first
sliding window to form a subsequent sliding window .

48. The method of claim 44 in which collecting
additional fractions of said data points comprises removing

10 at least the earliest point of data collected in said first
sliding window and adding to the remaining points of data
at least one new point of data to form a subsequent sliding
window.

49. The method of claim 44 which further
15 comprises designating one of said preselected fractions as

forming a basis set of data points and determining its
characteristics . v

50. The method of claim 4 9 which further
comprises comparing the characteristics of said basis set

20 of data points with the characteristics of the remaining
preselected fractions.

51. The method of claim 50 in which one of said
characteristics includes the slope of each preselected
fraction in its respective sliding window.

25 52. The method of claim 50 which further

comprises detecting the presence of aberrant data points on
the basis of such comparisons.

53. The method of claim 52 which further
comprises counteracting the deleterious effects of said
30 aberrant data points if the presence of such points has
been detected.

WO 92/01928

- 81 -

PCT/US91/05071

54. The method of claim 53 in which
counteracting the deleterious effects of said aberrant data
points comprises performing a step selected from the group
consisting of utilizing median filter techniques,

5 eliminating the data points which lie outside the desired
range and interpolating the remaining data points/
utilizing digital filter techniques or discarding the
sliding window which contains said aberrant data points.

55. The method of claim 1 which further
10 comprises comparing the slope of the data points in a

particular time window to a range of expected values to
determine the integrity of the corresponding signal
measurement.

56. The method of claim 1 which further
15 comprises comparing the difference between the slope of the

data points in said .first time window and the slope of the
data points in said second time window to a range of
preselected limiting values to determine the integrity of
said signal measurements.

20 57. The method of claim 1 which further

comprises comparing the mean value of the data points in a
particular time window to a range of expected values to
determine the integrity of said signal measurement.

58. The method of claim 1 which further
25 comprises comparing the deviation of the data points in a
particular time window from the linear interpolation of
said data points to a range of preselected limiting values
to determine the integrity of said signal measurement.

59. The method of claim 1, 2, 3, 4 or 5 in
which relating said signal measurements comprises:

WO 92/01928

- 82 -

PCT/US91/05071

(a) interpolating the data points in said first
time window;

(b) interpolating the data points in said
second time window;

5 (c) extrapolating said first time window

interpolation forward to a position located between said
first and second time windows;

(d) extrapolating said second time window
interpolation backward to said position; and

10 (e) calculating the ratio of the values

obtained from said extrapolations, from which ratio the
concentration of said preselected analyte species can be
determined.

60. The -inethod of claims 1, 2 or 3 in which
15 said calibrant fluid is comprised of an aqueous liquid or a

wet gas .

61. The method of claim 22 in which said
potential * changes are applied in the form of pulses,
incremental steps, sine waves, linear sweeps or

20 combinations thereof.

62. The method of claim l r 2, 3, 4 or 5 in
which at least two signal measurements are performed in
each of said first and second time windows, one of said
signal measurements being carried out at a first applied

25 potential and the other being carried out at a second
applied potential.

63. The method of claim 62 in which said signal
measurement includes measuring the current.

WO 92/01928

- 83 -

PCT/US91/05071

64. The method of claim 62 in which said first
and second applied potentials lie in the range of about 100
to about 300 mV.

65. The method of claim 62 in which one of said
signal measurements in said first time window is carried
out at an applied potential of about 125 mV and the other
measurement is carried out at about 225 mV.

66. The method of claim 65 in which one of said
signal measurements in said second time window is carried
out at an applied potential of about 225 mV and the other
measurement is carried out at about 125 mV.

67. The method of claim 62 which further
comprises determining the slope of the line defined by at
least two points on a signal versus applied potential

15 curve, which points are obtained from said signal
measurements performed at each of said time windows.

68. The method of claim 67 in which relating
said signal measurements involves comparing the slopes of
said lines' obtained from said time windows to determine the

20 concentration of said preselected anaiyte species.

WO 92/01928

PCT/US91/05071

WO 92/01928

PCIYUS91/05071

WO 92/01928

PCT/US91/05071

4/15

Millivolts

Ll.

WO 92/01928

PCT/US91/05071

WO 92/01928

PCT/US9I/05071

WO 92/01928

PCT/US91/05071

WO 92/01928

PCI7US91/05071

WO 92/01928

PCT/US91/05071

FIG. 4

WO 92/01928

PCT/US91/05071

WO 92/01928

PCI7US91/0507!

WO 92/01928

PCT/US91/05071

WO 92/01928

PCT/US91/05071

INTERNATIONAL SEARCH REPORT

International Application No. PCT /DS91 /05071

i. CLASSIFICATION OP 3U»JICT MATTtW (,l e.verai cla»„ncat,on »ymooi, apolyt , wcWcai » ttl) «

^WW^Wi^^WW^ 31/00 G01F 1/66
U.S.C1.: 204/400,402,406,153.1,424,433 73/861.28

H FIILOS SEARCHED

Clast>r. C j:.on S/stem

U.S.C1.
IPC (5)

Minimum Doeum«nt«tion Searched »

Cla*»ificat«on Symbols

204/400,402,406,153.1,424,433 73/861.28

COIN 27/30,27/54,27,00,31/12,27/02,27/46,27/28 G01P/1/66

Documentation Searched other than Minimum Documentation
to tho Extent that such Document! art Included in the Fields Searched *

III. DOCUMENT* CONSIDERED TO BI RELEVANT •

Cettaory

X
Y
Y
Y
Y
Y
Y
Y

Citation of Document. " with indication, where appropriate, of the relevant passages 9

US, A, 4,225,410 (PACE) 30 September 1980 *
(See entire document),

US, A, 4,935,106 (LISTQN ET AL.) 19 June 1990
(See entire document).

US, A, 4,935,105 (GHURGHHOUSE) 19 June 1990
(See entire document).

US, A, 4,686,479 (YOUNG EI AL.) 11 August 1987
(See entire document).

US, A, 4,293,522 (WINKLER) 06 December 1984
(See entire docunent).

US, A, 4,033,830 (FLETCHER, III) 05 July 1977
(See entire document).

US, A, 4,535,285 (EVANS ET AL.) 13 August 1985
(See entire document).

US, A, 4,019,966 (REMES ET AL.) 26 April 1977
(See entire document).

Relevent to Claim No. *

1-7, 9-11
1-12, 15, 16
1-5, 21,22,34
10-18, 60
19, 20

22- 26, 38

23- 29,32,34-37
38

* Special ca tegories of cited documents: «
"A" document defining tho oonerei state of the art which la not
considered to be) of particular relevance

"V earlier document but pubtJehed on or after tho international
filing data

document which may threw doubts on priority doimU) er
which if cited to estebiishthe puhhcation date of another
citation or other apecia! reeeon (at opacified)
-O* document referring to an oral disclosure, use, exhibition er

"V later document published after tho International flUng date
or priority date and not hi conflict with the application but
cited to understand the principle or theory underlying tho
invention

M X" document of particufar relevance; the claimed invention
cannot be considered novel or cannot be considered to

T* document of parpcuiar _ . . _
cannot be considered to invoke
lit combined with

»P- document published prior to the intarnational filing oat* but
later than the priority date claimed

in tho art
"4* document m ember of the

... the

an mventrie step when the
> or more other such doce-
obvioos to « person skated

iv. ctirrtncATiON

Date of the Actual Completion of

16 October 1991

the International Search

Oats of Mailing of this International Search Report

25 NOV 1991 /

ol Auttwtttad Oflfetr

ISA/US

International Application No/ pgj /US91/05071

FURTHER INFORMATION CONTINUE* FROM THE SECOND SHEET

US, A, 3,672,843 (ROSSE EI AL.) 27 June 1972
(Sea entire document).

US, A, 4,787,252 (JACOSSON ET AL. ) 29 November 1988
(See entire docisnent).

US, A, 4,464,230 (UHK3)0N) 07 August 1984
(See entire document) .

US, A, 4,897,162 (LEWANDOWSKI ET AL.) 30 January
( See entire document ) . 1990

39, 41
43-59,62
61, 63

OBSERVATIONS WHERE CERTAIN CLAIMS WERE FOUND UNSEARCHABLE 1

This international search report has not botn established la respect of certain claims under Article 17(2) (a) for the following reasons:
t. □ Claim numbers . because they relata to subject matter M not required to be searched by this Authority, namely:

2-0 C 1 *'" 1 numbers ...

ments to such an extent

because they relate to parts of the international application that do not comply with the prescribed require-
that no meaningful international search can be carried out », specifically:

xQ ClaJm numbers^

, be*«e^sy»depe*de^

PCT Rule 6.4(a).

VI. Q OBSERVATIONS WHERE UNITY OW INVENTION IS LACKING »

This Internationai Sesrchlno Authority found multipJo inventions In this International application as follows:

.1.Q As all required additional search fees were timely paid by the applicant, this international search report covers all searchable claims
of the international application. '

MD As only soma of the required additional search fees were timer, paid by the applicant this international search report covers only
those claims of the Intsmstionsl application for which fees wars paid, spectflcsfly dalmss

3.Q No required additional search fees were time* paid by the applicant Consequently, this International search report Is restricted 1
the invention first mentioned in the claims; a Is covered by claim numbers:

be searched without effort fustifying an additional left, the International Searching Authority did not

4.n As aUsearebableclaims could

UJ invite payment of any additional fee.

Q The additional search lees were
□ No protest accompanied the payment of
_______

accompanied by applicant's protest

I "»»™««on»l Application N».~PCT7US91 /05071

J ^DOCUWINTS COH8.DIW.P TO £ HEL.V ANT (COWTIHU1D ««0, THE SECOND SHIT,

C *" g ° ry ' I """O" »' Pocunwnt, ■» with IndlcUon, wh»r« .pprapriw,. of »» rrtwinl

US, A, 4^86,590 (TITTLE) 12 DECEMBER 1989
(See entire document).

US, A, 3,765,841 (PAULSEN ET AL.)

16 OCTOBER 1973 (See entire document).

R»!*r»nt to Claim No »

PCT/tSA/210 (extra *h*tt) (May 19S8)

THIS PAGE BLANK (uspto)

Internet Archive Audio

Featured

Top

Images

Featured

Top

Software

Featured

Top

Books

Featured

Top

Video

Featured

Top

Mobile Apps

Browser Extensions

Archive-It Subscription

Save Page Now

Full text of "USPTO Patents Application 10687850"

See other formats